Techstrong TV July 24, 2025

Hey everyone. It's the C-Suites dilemma, AI cut bait or double down. You're watching Techron Game.

Hello everyone. Happy Thursday. Welcome back here to Textron Gang.

We've got a great lineup of stuff to talk about, topics to talk about here on the gang, and we've got some really, really interesting, fun people to talk about it with. A lot of 'em, of course, have already been here before. Uh, let me intro quickly introduce our gang members.

Today we have Terry Robinson, Garima, Bal, guy Courier, and you have Geni Karaman, of course, our Dean Mike Ard, gang members. Welcome Mike. You know what, what is this poor old C-suite to do?

They're dropping like flies. Should they, should they double down on ai cut bait and run? What?

What's the deal here? It seems like there's this kinda hope in the C-Suite at least that ai, somehow or other, is gonna change the economics of doing business. And, and they've all seemed to grasp it.

It's kind of becoming something of a religious moment. But we can go after the average worker and including a survey also from other senior leaders for that matter that PWC just did. It suggested we don't really trust AI agents for anything mission critical.

We're using them for support functions. And that seems akay, but that doesn't add up to math. That says that the economics of the business is gonna fundamentally change.

And I know you have a post over on Techstrong ai Alan talking about that same issue. So, I mean, what is your real assessment of what's going on here? Because I think we're setting ourselves up for a bit of a fall.

Yes and no. Yes and no, right? Because, you know, if, if you are A-C-E-O-C-T-O-C-P-O, whatever, you're c-level leader of your organization, you have a fiduciary duty almost to use what's out there to make your organization the best it can be, as profitable as it can be, as successful as it can be, whatever the corporation's goals are.

The organization's goals are, r AI is certainly a tool that can help. It could, it could save overhead, it could reduce labor costs. It can make you more efficient.

It could automate more. It could do make you better. On the other hand, as we've seen in stories coming out this week, it could delete your production database, cover it up and make up information too.

So, and when it does that, you know, blame the poor ai or who gets the blame? Well, ultimately the buck stops here at the C level desk. And so if you're a C level, the dilemma here is, do I, do I go all in, put my chips in the middle and say ai, damn, you know, damn the torpedoes full speed ahead.

Or do I say maybe risk being a laggard. Go slow old school and let someone else fall on their sword before I try to walk through this minefield. Now, the article, we, we we're citing here up on Techstrong AI is one I wrote, and I, I try to give a blueprint, a roadmap for C-level people to, to kind of look at and, you know, make Nexus decision points as they roll through this.

But, you know, uh, there's another article on, on the Textron ai well as well that, you know, according to a p WC survey, they're not giving, uh, trust. They're not trusting AI and AI agents in high stakes mission critical use cases just yet. That might be prudent given news this week.

But, um, but the, you know, when you're a C-level guy, either you're going for it or you're not cautious people. I mean, you know, cautious CEOs who lay back and let others go first generally are not, are not the, uh, you know, they don't have a long lifespan. Most companies wanna seize the moment.

So I, I don't know what the right answer is. My tendency full speed ahead, damn, the torpedoes, right? If you're gonna go down, go down fighting.

Rob, I think we got a little break there, hopefully, but Guy, Go ahead, Guy, you wanna jump in here for a second? And, you know, what's your assessment of what's going on here from where you, where you sit? ai.

Um, uh, it's called Cut Beit or Double Down. I really recommend the article that he wrote published yesterday about this, because it is quite practical and it names five ways that, uh, the C-suite or management can, uh, make sure to, um, to approach and adopt and use AI effectively. I do think one thing continues to be missed or misunderstood, um, and has been since the beginning.

Um, and if, if you forgive me, let me just say that it's sort of human nature to, to run with anecdotal evidence instead of, you know, uh, uh, what, uh, you know, um, objective or empirical or whatever. Like, there's a lot of words for trying to understand in the aggregate or statistically speaking, what's going on in this case, what the benefits of AI are. Um, everyone's personal experience with AI is something that used to take me an hour or two hours.

Now it took me five minutes, whatever it is, looking stuff up on the web, drafting an article, whatever it is. So the, uh, you know, how many folks on this panel here in this show, um, think that or believe that the market thinks that the main benefit from AI is productivity, that one worker's gonna be able to do what two used to do or more, or that, uh, each worker's gonna be able to produce more. That is the general consensus, and I have been maintaining since the beginning.

But that is not the benefit or the metric you should look at. The two you should look at are, uh, quality and reliability. Quality sounds dumb, right?

Because AI makes mistakes, hallucinates, and a lot of AI is actually getting worse at the moment, rather than getting, getting better in terms of accuracy, let's say. But quality goes up when you have a continuous companion, whether automated as an agent or at your command in a chat or something. When you have a companion, even an artificial one to help you do your work, it's still your work.

And you can go through more review cycles and add things that you might not have thought of and construct things in ways you might not have thought of thanks to your AI companions. So the quality of your work goes up, the reliability goes up, because all those things that you have to do, like research or whatever, that you sit there putting off and, oh, maybe I'll walk the dog now, or maybe this is the perfect time for me to clean the kitchen. Instead, you have your buddy, your enthusiastic buddies who just go and do it for you.

If CXOs, if, if CXOs or business leaders are thinking in these terms, not in terms of, oh, now I get to fire half the marketing department, but rather I can get better results and I can measure those results, that I think is the, the, with Alan with great respect to what you wrote. 'cause I think it's really useful. I think that's the one thing you were missing.

What are you using this for? Not, not to you, you, you tell people, you, you're providing the, the guidance to identify that that's true. But we are sort of lacking in general in the market with nice, strong points of view.

Like I just gave, and you can disagree with me, but it's a pretty strong point of view as to what you should be using all of it for. And I would say for ai, it is quality of work and reliability that the work will get done on time as well, and productivity flows from that. Evgeni, you raised your Head.

Yep. Yeah. So guys, thank you very much.

I think it's a very important point. I want to add a couple of ideas here and maybe kind of have a small debate. 'cause you mentioned helper.

When I think about a helper help is somebody helps me. So basically, each of us now have a helper, or two helpers or three helpers. But in the environment, when we are going in a manufacturing or hospitality or anywhere else, we're talking about C-suite, I'm thinking about a mechanism, an AI that help my customers or help somebody externally.

So it need to be doing multiple things. This could be customer success. This could be providing advice, this be potentially booking flights, constant flight, whatever it is.

And what, where my mind is going is not one or zero working, not working is what's happening in between what happened when the AI make a mistake or we found a bug. Because if it, it's a thing. The AI is a human, or like, as a human, it make, make mistakes.

But when a human making mistakes, it's change. When AI make mistakes, it's working at a hundred times thousand times faster. So the mistake can escalate much faster.

How do we debug this? How do we create this? How do we have high availability?

How do we fix the car on a fly while we're driving? Do we have another LLM AI that we take this out and put to somebody else? What do we fire our entire booking department for travel and put the ai and now we need to fix it.

So let's take that example, let's take that example. It's a very good perspective. A, I mean, thank you.

But let's just take that one example of you're right in the case of, uh, a customer service, a chat, or whatever it is, you're, you are providing that buddy, i, i described to the customer, if your goal in use in creating that aspect to your interactive application with the customer is to help the customer get a higher quality result more reliably, then you'll develop that AI in, in a better way. In my opinion, in my opinionated opinion, the usual goal for putting AI in those scenarios is that you can have fewer employees on your side. It's your benefit, not the service benefit.

And that's a fundamental problem. And, and it affects how you measure ROI, Alan, which is your number one thing to make sure of, and I completely agree. How are you measuring that?

ROI, that ROI may not be in terms of lesser salary or what have you. It could have to do with like a better brand, better perceptions, higher, uh, a net promoter score, like that sort of thing. Agreed.

Agreed. Well, look, I it's not though it's a a, an exciting time to be a C-level person and what AI can do for your business. This ain't a slam dunk, right?

This is not just full speed ahead, this real right? This is a time, it's a little choppy. The waters out there right now.

And, and, and I, you know, a little caution may go a long way though. You don't want to be the laggard, I think, I think there's a, there's a lot of room for fact checking in the world. We need to hire fact checkers now for Ai.

Let AI fact check, let the AI fact check the ai. Hey, we're gonna take a break here on Textron Gang. Let's come back to our B block today, which is about platform engineering.

Nice topic to talk about. You're watching Textron Gang Discover Textron Group, the epicenter of tech innovation. We are your go-to for reaching IT leaders and practitioners worldwide.

Our secret impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients.

Let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Techron Group. Hey folks, it's Alan said, we're gonna talk about platform engineering.

com. We invite you to check that out. But it says that sometime after the turn of the decade, this platform engineering market will be worth, I don't know, I think it was $40 billion.

Is that the number, Alan, right? But Garima, let's go to you. I mean, here's my problem with this whole report, and I, it's my problem with any of these reports, but, um, so we were investing in other things in IT before platform engineering came around.

So we just double counting here because some of this was our DevOps spending and our IT infrastructure spending already, or is this actually on something called platform engineering that's net new, above and beyond what we were already spending? Thank you for that question. And again, um, I'll start from basics and then we go onto this question.

So I was reading this report as well, and there is another, um, uh, report from, uh, Gartner, which says that in 2026, around 80% of large software engineering organizations is expected to have a dedicated platform engineering team. And, uh, if you look back like this moment of somewhere started in 20 20, 20 21, and then, uh, it has picked up from there. But what is, uh, uh, to your question, what is fueling this growth, right?

I mean, of course, uh, there's a plethora of tools and applications, but I will actually, uh, uh, identify two, uh, two areas which is fueling this growth and it becomes more exciting for investment per, uh, per se from DevOps and SSRE practitioners. The first aspect is, uh, AI and the interest of AI native workloads, integration of AI and advanced technologies. And when you talk about commercialization of generative AI technology and other advanced analytics, this is pushing enterprises to upgrade their infrastructure, for example, including data structure centers and networking storage.

And I'll come to some of the recent announcements which will prove this point. Um, another, uh, growth, uh, example, and this is again, uh, based on, uh, you know, my findings is there's a substantial amount of interest in vertical powerhouses like banking and financial domain when it comes to platform engineering. So we have often talked about, you know, how lucrative it looks, uh, from large or enterprise level, this platform engineering, because it, of course, it cut downs the overload.

Um, it also has, uh, streamlined developer, uh, workflows. It has substantially contributed to developer productivity. But there was one thing which I was watching out for, what is it in for small companies or mid-size companies?

Because it was not proven that platform engineering is, uh, providing benefits because it becomes a limitation to a certain extent if you, uh, look at small companies, uh, who have to invest four, five people into, uh, upkeeping the platform. But now there's an interesting development because of this AI integration or AI native development. And this in, uh, interesting development, I will quote some of these cloud providers.

So, uh, cloud providers have jumped into the bandwagon, right? And the platform engineering has become an essential component for cloud service providers. For example, if you see what Azure has done, they have launched this dev box, which is basically, uh, ready to code environment, which is cloud based development environment.

And it is primarily targeting like small companies, right? So this is an, uh, substantial amount of kind of, uh, in, uh, has gained substantial amount of interest from small, uh, companies and even solar printers. Uh, another, uh, interesting development which has happened from a cloud perspective is let's say Azure arc.

If you look at Azure arc, what they have done is that they are not talking about interoperability the same game which they have done when they explored like beyond Microsoft, right? So Azure ARC is exploring multi, uh, cloud environment and how, uh, this platform would enable multi-cloud environment. Of course, they, uh, see the need in the AI era, right?

Um, if you, uh, look at other cloud providers, they are still kind of, uh, pivoting, uh, carefully. For example, Google has this, uh, GKE first, um, perspective on their development platforms as well as, uh, AWS, uh, it's also trading cautiously. But what the most important development of the week, and this is, was also covered by Techstrong TV, was Broadcom and the announcement, uh, by vm, uh, like Broadcom, that VMware, this platform, VVCF nine.

And what VCF nine is kind of doing is that it's enabling platform engineering concepts, uh, uh, to a certain extent for private cloud, right? So then it's changing the game a little bit here and Hawaii again, why this whole interesting development is happening is to a certain extent, it's fueled by AI and AI native software development. We can talk a little bit, uh, more about MCP, for example, and MCP marketplace, which is another exciting development which is happening.

And I do believe that MCP marketplace will be a new feature in platform engineering or platform engineering space. So it has a lot of potential in that, uh, regards, but I will give it back, Mike, to you for more perspective on this. So, so yeah, I, here's my take on this, guys.

Mike, to your original question, Greenman was a great answer. Platform engineering needs a Statue of Liberty. And since we don't seem to believe what's written under the Statue of Liberty anymore in this country, maybe we should give it to platform engineering.

And it says, send me your, what does the Statue of Liberty say? Send me your homeless. Send me your, you know, poor people send me poor your huddle mass, right?

Yeah. Your huddled masses, all that stuff. The deal with platform engineering is, it's not nearly new.

When you look under the covers, the name platform engineering is sort of new. And the, the title a platform engineer is sort of new, but the, the, the things that platform engineers do is not new. We've been doing them for years.

We had CIS admins doing them, and we had, you know, ops people doing them and what SREs do today and all, you know, so when we look at that $40 billion, a lot of it was money that we were already spending. Maybe it had, it was under different tents or different titles, and we're now pulling it under the umbrella of platform engineering as it pulls all these disparate un huddled masses of people and gives them a place platform engineering. So are you saying, Alan, that the, that the growth projections and, and total available market refers to the total available market for a marketing term, a new marketing term platform Engineering, not a, not a new marketing term would be aing term, but it's a new, it's a new grouping of existing dollars.

And, and when you look at the history of modern platform engineering, you know, since we coined the term, well, originally it really referred to can I manage Kubernetes? 'cause it was really tied into Cloud native and Kubernetes. But, but it's, it's morphed a little since then, it's grown.

It's not only now can I manage Kubernetes, but what's really driving platform engineering today is I need to manage my i internal development platform, internal developer platform, my IDP and the IDP is what's driving our platform engineering team. And, and that lends itself, gima, to your point, that lends itself to large organizations where you need an IDP 'cause you've got dozens, hundreds, thousands of developers. If you are a little shop with a half a dozen people coding, you probably really don't need an IDP.

And, and then hence, you maybe don't need a platform engineering team. So if you are an insurance company, okay, you are only doing insurance, are you a platform or you still have one function? If you are a bank, you may become a platform because you may do multiple things.

If you are a cybersecurity vendor like Palo Alto or McAfee used to be, when you have multiple products, you have a platform. So I guess we need to also think about the idea that if I don't have a very big development shop, as you're saying, or many different products, I'm not gonna be a platform then, then no, you, If you don't have a big development shop, you may not need an IDP and hence you may not need platform engineering. Is, is, I'm, I'm gonna go a step further though, because I think that this is a, the beginning of something larger and great that we're talking about IDPs and centralizing application development.

But let's be honest, the management of it and the enterprise is a frigging mess and has been four years. We have all these different silos. We spend a fortune on the cost of labor to manage all that stuff.

And mainly because we're spending a huge amount of time trying to integrate and maintain this stuff, it is fundamentally economically insane. Platform engineering is a step towards centralizing that in a way that makes it accessible, right? I have the notion of self-service, so people can take care of what they need.

And I'm not beholden to some autocratic CIO somewhere to do every stupid little thing that I want to do, but we have to get back to some fundamentals of the economics of it, which right now is kind of a disaster. So this is a step in the right direction, Right? I agree with Mike and Alan, both of you, because I, I wanted to kind of, uh, also shed some light on, uh, you know, why platform engineering becomes very important for DevOps practitioners, because history reminds us why DevOps was, uh, important for us.

We were lean, right? And with the plethora of tools, which we see, the complexity has, uh, uh, increased tenfold, right? So platform engineering is streamline those, uh, developer workflows.

This is like providing some kind of, uh, value for money for companies and investments, even if it was investments in 10 different directions. I think platform engineering is giving that lever or tool to ensure that the return of investment is secured by, you know, investing in the right kind of tools. I'm also following some of the talks.

One of the great talks this, uh, week was on DevOps, uh, Munich. Uh, there was a conference there where they were talking about graveyards of tools because there is a lot of tools which are being ob becoming obsolete. So, I mean, there, there is a potential of looking at it, uh, from a platform perspective, uh, very different, uh, different with a different lens.

And to your point, actually, you mentioned that, uh, if you're an insurance company or a banking company, I would say that every company is in a software company today. So this is very, very important that we, uh, ensure that we take our software workload developer experience very seriously. And why it is important, essentially becoming important for smaller companies is because, uh, it provides you the capability to exponentially scale, Right?

And scale is what it's about. And scale is what platform engineering's about. But hey, we gotta take a break.

We, we ran outta time for this topic today, I apologize. But we are gonna come back. We've got some cyber crime investigations, and we got a couple of cybercrime sleuths right here.

com is the leading resource for news analysis and education on challenges facing the cybersecurity industry. com covers all aspects of cybersecurity, including data security, DevSecOps, cloud security, application security, network security, security threats, and more. com has the largest selection of security content featuring breaking news, blog posts, podcasts, and more.

com to learn more. com. Home of security bloggers network.

Hey, folks, we're gonna shift gears a little bit and talk about cybersecurity because, well, it looks like the bad guys have figured out how to hack into the surveillance state and are using it to well commit murder. But Terry, you wrote this story for us in Security Boulevard, and is, is this kinda like some weird way of where the infrastructure that we created is now being hacked into by the bad guys for illicit purposes? Well, yes, and, and not surprisingly, right?

We've talked about this before, how the bad guys take the good stuff and, and use it as, uh, for their own nefarious, uh, actions. Um, so, so what, basically this Happened, they're like committing murder, by the way, like literally This time literally committing murder. We're not being ative.

Yeah, Yeah. The way, Mike, the way you said that it was like committing murder, you know, like, uh, like, like the kind of metaphor my mother used to use. No, we're talking about literal murder.

Sorry, Terry, just This Yeah. No, no, no, that's okay. This is, you're, you're right.

It's literally murder. Um, so what happened here is, uh, it, so this all stems actually from an IG report, uh, for the FBI that came out, uh, last month, that they're just, they're actually trying to, uh, audit, like, uh, the FBI's efforts to mitigate the effects, um, of technical surveillance, which is just ubiquitous now, right? It's everywhere, not just in Singapore, which we've, I think we've discussed Singapore, how creepy it seems sometimes that these cameras are all over the place, but they're, they're everywhere.

And, um, one of the, the stories that they recounted in this audit was, um, about in Mexico City, um, there was a hacker affiliated with, uh, El Chapo's cartel, or one of his cartels. And, um, this person or group or whatever, um, actually surveilled the FBI, uh, building there, or the embassy in, uh, Mexico City. And they identified like an attache and, uh, other, other people.

And then they focused in on this, this one person, and w they were able to access his or her phone records, location data, contacts, all of that. And then they used the, the web of cameras in the city to trace, um, informants, like track them, and then in some cases kill them. Uh, other cases I think they probably just put pressure on them or whatever.

But yeah, so literal murder, um, and it was a, like, sort of this interesting combination of tech. And then that old school kind of, you know, you used to see in the Rockford files or something where somebody, you know, is, is uh, is, uh, casing a building, uh, and, and watching people, you know, like literally a person out there observing people. Now they're not a hundred percent sure that this hacker was a person sitting out in, in front of this building watching people go in.

Um, there could have been all sorts of, um, technology involved in that. But, um, so First of all, I thought they fired all the, uh, uh, uh, inspector generals. Well, so, and so that's like, you know, when I was first looking into this, um, I had a little moment of satisfaction.

Those igs are really super valuable to government and the fact that we got Rid of, so if you don't fire them, they are, If you don't fire them, the fact that we got rid of, of so many of them earlier this year makes me even more concerned. Now, I mean, about what I mean with this report, I, I, I, you could probably count his employment on two hits, but, but, but there's another thing here. Look, the real issue here is in a surveillance state, the, the room for abuse, whether it's by the government itself or people who are hacking into leaky government, surveillance infrastructure is great too.

Great, too great. And the, and sometimes you gotta weigh what is the, the, the advantage to the, to the downside of these things and, and make a determination if we're going to have this type of surveillance capability, number one, we've gotta make sure that it legally by government sources, has to be extremely supervised and qualified. Number two, what Kind of good luck was that, though?

I don't think we're heading in that direction, And that's why we don't have igs the way we used to. But number two, it's ludicrous to think that we would do such a shoddy job of setting it up up that would, would allow the, the quote unquote bad guys. And who's a bad guy today?

It's hard to tell when they're all wearing masks, but mm-hmm. It would allow the bad guys to, to harness the same, uh, infrastructure that we've set up for this. So it's kind of a double whammy, which leads me to believe we shouldn't be doing this stuff, right, until we have guardrails in place.

I think it points to what, how security, at least previously, and hopefully this is getting better, although I can't say for sure, was always like an afterthought when people were setting up or groups were setting up, you know, things like this, right? They didn't really think too hard about security. Well, I mean, that's my own opinion that I don't think they considered all the options.

I also think that the bad guys have gotten so far ahead of, you know, uh, the good guys in terms of how they use technology and everything. But it is a, you know, it is a, a a hard to win battle when You're on the, depends who you think of as the bad guys and the good guys. Like I said, they all wear masks.

There's the ambiguity there is, you know, who are the bad guys? Who are the good guys? It gives me great comfort that the United States Government awards these contracts to the lowest cost of partners with the least amount of cyber cybersecurity expertise.

So I think it's gonna be great. I know, well, merit, you know, we're on merit now here, so, um, good to, always good to see. But, um, yeah, so that all that's of a concern.

I will say right here, one of the things that they, they, they do think that maybe there were some insiders, uh, within the agency that possibly shocking. Yeah, shocking going on here. That's what struck me, Terry, was, was just as we harp on, you know, uh, AI needs human intervention, cyber needs, human intervention, but it also needs to address the human factor.

And, and it seemed to me that hacking into a video surveillance system is one thing. A lot of these are pretty old. And, you know, the, the cloak and dagger of somebody going over to a line and tapping it, like you saw in Oceans 11, which is pretty much mostly fake, but could actually occur in, in, in the case of video surveillance, you could see all that.

But it seems critical that they use the same old, usual, you know, pay somebody off method too. Could be, you know, it's, it's a, it's a twisted world, a twisted world. So the IG report also did give some, you know, they, they did offer a few recommendations, although honestly, they're fairly basic, so I'm not sure how much help they'll be, but it was good to see the IG active and getting out there on what is an important topic.

Absolutely. Alright guys, I think that wraps up our textual gang for Thursday. We hope you, you enjoyed it.

As usual, we have Textron TV following this, um, on our network. You can also check this out on our Textron tv, YouTube channel, our Textron TV webpage, as well as our Textron TV OTT app, which is available on Apple and Google Play and Apple TV and Roku TV and Amazon Fire. So no matter what you watch on, we've got a text on TV for you, uh, yang members, thank you so much for a lively, great conversation.

Mike. Thanks for leading it. We'll be back tomorrow with more, but until then, this Alan Shimel we're out.

Hey everyone, it's ama welcome back here to Tech Drunk tv. Let me introduce you to our next guest. Her name is Nicola acu.

Yes, we got it. I think we got it right. Nicola is the Chief Sustainability Officer at NetApp, and let's welcome her here to Tech Drunk tv.

Nicola, it's very nice to meet you. Thank you for coming on. Um, I always like to give the audience a sense of who they're talking to, and we don't get to talk to a lot of chief sustainability officers, to tell you the truth.

I could probably count on two hands how many, unfortunately. So tell us a little bit about you, your background, and how you came to be the, uh, chief Sustainability Officer there at NetApp. Great, well, it's a pleasure to be here, Alan, and, um, delightful to share this conversation with you.

And I think we're gonna change that. I think in the future, you're gonna be talking to certainly a lot more chief sustainability officers and, uh, certainly heads of it, sustainability. That is something, um, I think we're gonna see in the future, but we can talk about that later.

Um, just briefly, who I am, um, I am newly appointed, well, actually not newly appointed. I've been in the role for a year. It feels like newly appointed because Yes.

Yeah, Time flies, right? When you're having fun. So I've been at NetApp for a year, and I come into this role as a career sustainability professional.

I, uh, grew up in tech. Um, prior to NetApp I spent 14 years at VMware. Hmm, very cool, very cool.

In a sustainability role, Correct? In a, uh, social impact, sustainability, innovation, and, um, yeah, or bringing together the, uh, environmental, the social and the governance responsibilities of a company. So, Nicola, don't be insulted, but I'm sure there are people out here who are asking themselves, well, what exactly does a chief sustainability officer do?

Yeah. How would you answer them? It's a great question.

And a chief sustainability officer's role is first and foremost to understand the broader environment within which a company and business is operating, to interpret that external environment for risks and opportunities related to environmental social governance issues. So our job is to interpret the signals from the markets and from the broader environment and understand what that means, and put together a strategy for the company to mitigate risk and optimize opportunity. So in the case of my role at NetApp, I like to describe it as sort of the three, three pillars of, um, the portfolio.

One is to identify what those risks and opportunities are and to focus on what the most important ones. So for a company like NetApp in the business of data management, the top issues are energy efficiency and circularity. We can talk about what that means, but in essence, it is the, um, where the physical components of our infrastructure come from, where we procure them from, how we build them, and what the end of life is.

So that's number one of the portfolio set strategy by understanding the external environment. Number two is I protect the business by interpreting reg regulation and, uh, uh, uh, policy and emerging changes in the regulatory environment around us. So I make sure that the company is regulation ready, um, and we're in compliance with, uh, appropriate environmental social governance laws.

And then the third component of my portfolio, which to me is the most exciting and interesting. And really what, what gets me up every day, um, is the innovation aspect. So based on those other two parts of my job, I serve as a business partner to our engineering and our product groups to interpret that information and to help build better products and solutions for our customers.

I love it. Thank you for that. Um, you know, in the same vein, people are saying, okay, so NetApp must be very committed to sustainability 'cause they went and what you are a year ago as chief sustainability officer, excuse me, where does sort of rubber meet the road in terms of sustainability at NetApp?

Look, sustain. While I am new to the company, sustainability is not new to the company. Um, NetApp is a company that's had a really excellent reputation for doing the right thing for a very long time.

Um, so what is new, however, is the changing environment around us. So from a regulation perspective, right? Um, being ready to address different jurisdictions around the world where we operate, um, is all about actually change management and implementing systems and processes to gather the necessary data to report on sustainability issues.

So that was one driver. Another driver is frankly, commercial. In the places around the world where we operate, sustainability is increasingly a requirement of doing business.

And so having a competitive strategy in the market around how our products can better enable sustainability outcomes for our customers was an important factor that that is, um, that is undeniable. And then, you know, the, the third component is about what I call future proofing. Looking at what's coming and recognizing the need to build strategy around sustainability, around data efficiency, around energy efficiency, around resilience, into our broader sweat suite of offerings and into our vision for the intelligent data infrastructure of the future.

Love it. Thank you again. Alright, let's shift gears a little bit and talk about our topic of discussion today.

You know, SoftBank has pledged a hundred billion dollars. This one's giving $10 billion. That one's giving $600 million.

That country is building, you know, a multi-billion dollar AI factory. You know, before you know it, we're talking about real money here as the, the joke goes, right? But I, I did, you know, all kidding aside, I did a, a segment on Textron gang on this about two or three months ago.

There's been north of one like trillion dollars pledged for AI data centers, AI factories as we're calling them. Um, over the last year, how much of it will become real, I guess remains to be seen, but certainly the pledges have been made here. Now the fact of the matter is, I don't know if we generate enough electricity in the world today to power all of those AI factories and to do so we'll bring untold damage over our heads, onto our heads from an environmental perspective.

So if we are gonna achieve our goals in ai mm-hmm. It seems to me we've got to build sustainability into it. Sustainability has to kind of go hand in hand if we're going to scale those heights.

Um, wondering what, you know, as a chief sustainability officer, someone who's probably much closer to this than me, what do you think, Alan? I don't think there could be a more pressing conversation than that right now. Look, the, the promise of AI is transformative, and that is why so much investment is going into it right now.

Um, but in order for us to unlock that promise, we have to get smarter about how we're gonna get to the future. Um, and you know, I I like to say that sustainable AI starts with data, right? Absolutely.

We've gotta understand the, the data that's coming in the pipeline, right, that is driving this massive build out. Um, that is number one, right? And then beyond that, obviously the broader ecosystem of understanding where those electrons are coming from that are going into the systems and the photons that are coming out on the other side, right?

Um, so, so let's just, let's just talk for a second about data. And, you know, last year, uh, the, uh, IT sector, um, you know, the, the, the consumption of power, um, and energy is roughly the same of the, as the air, the airline industry, which is about 2%. Uh, which is kind of mind boggling, right?

That AI and that, sorry, that data, data companies and data centers can consume 2% of, of the global energy, right? By 2030. And the, and look, the, the, the numbers keep shifting, right?

But we know that they're shifting in one direction, and that's a hockey stick direction that by 2030 it'll be between 10 and 20%. Now, that's extraordinary. Um, and when you look at the, the, you know, the, the, the challenge on the, the front end, right?

The data piece, um, there's an an extraordinary amount of that data that is just not useful, right? But we call this dark data, and that's anything between sometimes 20 to 40%, sometime in, in some research, it's over 60% of that data is created and never used again. So that doesn't sound like a really smart setup, right?

And what's, what's the, what's the older dodge about, uh, innovation, right? The mother of, of, of The necessity, Necessity is the mother, Is the mother of invention. Invention.

So we've got on one hand, um, this extraordinary promise and this massive growth of data that's fueling that is behind ai. On the other hand, we've got constraints on energy, right? There are parts of the world, in fact, I was just in Europe, um, uh, recently meeting with customers and partners, NetApp customers and partners.

And, um, I learned that there are cities in Europe that have put a moratorium on any new data centers because of the energy demand, right? And the co the seeming set the setup of conflict between data center needs and community needs. So you've got the setup, and in my view, and this may be a little bit of con controversial view, but in my view, something's gotta give.

And to me, that is where there's an opportunity for innovation number one on how we build out data centers, but, and number two, on data infrastructure itself. And that happens to be NetApps sweet spot, Right? So I don't work for NetApp, as you could tell, as you know.

And, and the way I look at it is, is it's really, um, you know, net zero or zero sum is there's only two things we could do here, and maybe we gotta do both. Number one, we've gotta generate more power, but that power has to be clean, sustainable, renewable, good energy, however you want to define that. Number two, we've gotta be more efficient about the energy we use.

And, and as you said, you know, the dark matter, the dark data issue, right? Is, man, if you could, you know, you're never going to get a hundred percent of that out of the equation, but even if you cut it in half, you're cutting maybe up to 30% of your, of your data storage out of the equation. Um, so we, we need to work on both of these fronts, right?

And we're seeing it, look, we're seeing it in a lot of the, the new data center build outs. We're seeing them build them near hydroelectric sources. We're seeing here in the us Microsoft, I don't know how smart this was, but Microsoft made a deal with the company who operates the old three mile island nuclear facility of the infamous three mile island.

You know, they're gonna re re uh, reenergize that and, and Microsoft's gonna build data centers near there, uh, whether it's geothermal or, or what have you. People are definite, or, you know, with a new generation of nuclear as well, these mini nuclear plants and stuff, I sort of, one comes into, I, I dunno if it was the thorium or whatever. Anyway, the, the bottom line is, is we're trying to be smarter about building data centers near energy sources that are renewable, cleaner, cheaper.

I don't know how much help NetApp does brings us there, right? That's not really your sweet spot. But being more efficient in our use of data center, uh, in our use of AI in the data center, in these AI factories as we're calling them, being more efficient there, especially when it comes to storage, is something that's right in the sweet spot, sweet spot for, for NetApp.

So what can we expect to see in that regard from NetApp over the next, I don't know, 18 months? I mean, it's hard to look beyond that. Yep.

Look, you know, Ellen, you, you're absolutely right in that there's, um, and in fact, I I I like to describe it as sort of two sides of, of the coin. There's one is the how do we make AI sustaina more sustainable? And then there is AI for sustainability.

So that's another conversation that we can have. Um, and as you rightly pointed out, NetApp is not in the business of building data centers, but as you, as you, as you outlined, right? There is a huge amount of the, you know, sort of the, I think of it as the, the race to the future around how do we, how do we build data centers that are smarter and that are AI first, right?

They're, they're built for the, for the, um, the, the capacity and the load that, um, that we need. So NetApp is sort of, kind of part of that middle layer and, um, and we think about, you know, data first and foremost. So that's why I always say sustainable.

It starts with data and understanding that data estate, um, is gonna be critical now and in the next 18 months. And today, you know, our, our approach to sustainable, I, uh, data infrastructure has four key elements or four building blocks, if you like to it, Alan. One is kind of Sesame Street simple store only the data that you need.

And n NAB has the capabilities to help our customers and partners, um, with things like deduplication, compression, compaction, you know, things that we should are familiar to your audience, right? Um, let's not underestimate the power of classification and tiering knowing what, what your data is, is critical to doing, taking the second step, right? And that is understanding the energy associated with that consumption.

And so today, NetApp has tools that our customers can turn on to get visibility observability into what's going on in the data estate. So, um, we have a dashboard that's built on, um, blue XP that provides real time insight into the energy load and therefore the carbon impact of their infrastructure. And what's beautiful about that is it's not just like a dashboard telling you how fast you're going, it's gives you feedback on actions that you can take to reduce that impact.

So of the customers that have turned on this feature, over 70% of them have taken actions to reduce their, uh, energy, their, their energy footprint, and therefore their carbon footprint. So, um, I think those two things are, are, are simple things you can do today to manage that massive pipeline that's coming in, that pipeline of data. Um, then the, that's, so that's building block one, building block two.

The second, uh, the third building block is to maximize that on-prem efficiency. So using the best hardware that is designed like NetApps is with a storage efficiency guarantee built in, um, that helps you be massively more efficient given the hardware choices that you make. And then the fourth building block here is to migrate where it makes sense in your infrastructure to migrate loads to different data centers with a different carbon profile based on your needs, whether that's, you know, cold storage or, or other, other solutions.

So in a nutshell, that's today what NetApp can do. Um, and so you're gonna see us spending more time, um, and, uh, really helping our customers understand that so that they can get a handle on sustainability today while building their infrastructure for the future. Excellent.

Nico, I have one more question because we're about outta time. Where can we stay ab abreast of what you and the NetApp folks are doing around sustainability? Well, uh, our website for sure.

Um, and you know, I, I share on LinkedIn quite often, um, what we're up to, that's probably the most, um, let's say up to date, um, where I share, you know, some of the stories around, um, the work that we're doing. And then of course we have an annual impact report, which will be coming out at the end of the year that really summarizes, pulls it all together, how NetApp thinks about both the, the data aspect of sustainability and then the operational aspect, which is the, um, the, the, the product itself and the components that go into making our, uh, world class storage systems. Love it.

Nicole, we're about out of time. I'm sorry, I gotta Well, but let, let's continue this conversation. You know, sustainability is something we've covered here consistently at Tech Trunk, so something that's important to us as it should be to everyone, keep up the great work, keep doing what you're doing, do come back and keep us informed.

Okay? I look forward to it. Alan, this is not going away.

This is a topic that is relevant to I agree almost everyone I talk to. So, um, It's relevant to us and our children. Until next time, Absolutely.

Thank you. Nicola Kott, chief Sustainability Officer NetApp here on Textron tv. We're gonna take a break, we'll be back in a moment.

ai Leadership Insight series. I'm your host, Mike Zora today with Anan Supra, who's CEO for zipper. And we're talking about how AI will be applied to field service tasks because, well, as it turns out, a lot of these folks are struggling to figure out where they're gonna get parts from because there's all these tariffs out there.

Hey, Anan, welcome to Shaw. Thank you so much. Thanks for having me.

We've seen tariffs being, uh, aimed at or threatened at least at everything from South Korea and Japan of late to China, to Europe and everywhere else. And of course, we get parts from these places. We are the challenges that field service folks are encountering.

And how can AI kind of help them navigate it? Absolutely. Um, most of the service businesses that we know of, they are dealing with a complex landscape, high operational cost, um, in terms of labor parts and fuel drive to a lot of these complexity.

But they're also dealing with the challenges of, um, lack of skilled workforce, labor shortage, and many other problems. Let's drill down into the problem of operational cost. And there are three dimensions to the operational cost.

The first one is parts. As you mentioned, tariff plays an important role in parts and there's a lot of confusion in the field service community now. Our goal as a platform provider is to make sure that businesses are able to leverage data and insights to drive actions.

And one of the things that we are doing with AI is to provide visibility to the business and wanting to make them proactive rather than reactive. Now, one way to deal with the high costs is to ensure that they have visibility and predictability into what kind of parts are required for what kinds of jobs that they perform. And these businesses, they perform a lot of jobs like installations, repair, maintenance and so forth.

And with the advancement of AI and leveraging data and insights, we are able to predict what does the future going to look like for these businesses in terms of, uh, getting the right inventory stock. We are moving from the mechanism of these businesses managing a large warehouse of stocks to getting to just in time stocks because as we all know that any inventory that is out there for more than a week, the price would have changed. So the businesses are dealing with this constant situation of inflation, high costs, and not able to predict how much of stocks that they would require in stock for them to operate efficiently.

So we're bringing in a lot of different aspects with AI and automations to make sure the business have the right visibility, they can predict what kind of parts are going to be required, and getting to adjust in time, kind of a model of the warehouse. Mm-hmm. How does that just in time model work?

Because we, I think we've all experienced some sort of field service technician showing up and not having the right part at the right time, and so then they can't do whatever it is they were asked to perform once they got there. And that's frustrating for all concern. So how do I kind of manage that just in time process in a way that ensures that, uh, the technicians have what they need when they need it?

Yeah, great question. So for us to get to a just in time kind of a model, it requires historical information about jobs, equipments, parts and assets. Now as a system provider, we have all these data in the system.

We are creating sophisticated models with AI and machine learning to make sure that all these historical data of parts, equipments, jobs, skills of workforce, customer information is back to this model. And when a new job is created, the job predicts what kind of parts and materials could be required for performing this job. Now, let's imagine that you have a refrigerator at your home, whatever make and model it has an issue and it is out of warranty.

You call the service provider, you create a request. Now the service provider needs to know a bunch of information. They need to know that who you are, you are Michael, you have this refrigerator, what is the brand, make and model which year?

And it also needs to know what kind of issues could occur. And when an issue occurs of a certain type, let's imagine that your filter is bad and the refrigerator is making noise. And when that happens, it needs to get a particular kind of a filter.

So using all these data in the system, we are able to predict that when the job occurs, like say Friday of this week, they would need the filter to be drop shipped to your location when the technician arrives. And it's a sequence of event that needs to happen. So we are moving from the mechanism of, uh, a business managing all these parts and stocks into their warehouse, to getting that right in time before the technician arrives to your location, getting that part shipped and ensuring that when the technician is there, the part is there and they're able to replace the part.

It has a lot of moving pieces. And with the advancement of AI and machine learning, now we are able to collate all these data together and, uh, create this experience for the business to improve efficiency, reduce cost, and offer that great experience that end customers are looking to have. In effect, I'm turning the shipping company then into my warehouse because they're basically stocking things for me and delivering them when I need them.

Is that right? Absolutely right. Yeah.

Um, it's been a long held dream, but rarely fulfilled where companies are hoping that someday field service technicians might actually upsell stuff. And will that become easier to do if they have a visibility into what's in the environment that they're heading into? So maybe they can bring things or recommend things to sell while they're there, because I think that's been a dream, but rarely, Absolutely.

Fantastic question. So we talk to a lot of CXOs and organizations, especially service organizations, and historically field service is considered as an operational overhead. There is this term field service as a cost center to the business.

And over the years, there is a dream vision for almost all CXOs to transform field service from a cost center into a profit center. And in middle of all these transition from cost center to profit center, the most important persona is the technicians on the field because they are the ambassadors for the business, the brand, and they are not just doing break fixes, repair and maintenance, they're also selling what the businesses call as subscriptions. These are annual maintenance subscriptions.

That is one way of empowering field service technicians to ensure that they are able to sell new business to their customers. Now, selling new business to customers, whether it is subscription or annual maintenance contract, it also requires a lot of knowledge about the customer, about what kind of equipment is there, whether this is the right customer to sell an annual maintenance contract or not. And this is only one half of the puzzle.

Now, let's assume that a technician has sold an annual maintenance contract. During the course of the maintenance contract. The business needs to have visibility to determine whether this contract is profitable or not.

If this is profitable, they are going to renew the contract with the customer, maybe at the same price point. But if the contract is not profitable, maybe because the equipment is, you know, so old that it requires frequent repairs, the business may not decide to renew the contract. Or if they decide to renew the contract, they're going to renew the contract at a higher price point.

So at this point in time, data and historical information will continue to play. A pivotal role for these businesses and companies like us are leveraging the advancement of ai, generative AI and machine learning, combining all these historical data and providing that kind of an experience so that when a technician reaches a customer's location, they are prompted that this could be a right fit customer to sell a maintenance contract or a proposal. Um, how hard is it to predict what the cost of something might be 30, 60 days out?

Because not all services are an emergency. And then the question becomes, well, is can I see that something might cost less than 30 days than it does today or conversely, can I get some insight that says it might cost more? Yeah.

So it's, it's been an ongoing challenge for businesses and it's not a new challenge. This problem existed in the past, but this problem exacerbated during the inflation businesses that I spoke with and I speak with thousands of these businesses, they did not want to have any parts in stock because of the fact that they did not know what the price is going to be the next day or the next several hours. So what we are seeing a trend or a pattern in these service organizations is to move from a time and material kind of a scenario to what they call as flat rate pricing.

And there are many, many providers of flat rate pricing tools, flat rate pricing applications that offer businesses an integrated approach to managing flat rate pricing based on whatever services that they are offering. And this service could include labor, cost, parts and material cost, any overhead cost for this business. But from their perspective, they're just providing a flat rate pricing to their customers instead of it being time and material, which is extremely hard to predict in this current environment.

Mm-hmm. Um, and in some ways I imagine they're playing triage then on the assumption that some days the pricing will favor them, and some days it may be a little narrow than they like, but at least they have, uh, a fighting chances with say For sure. And that's where the next topic comes in, which is understanding the nuances of job profitability and costing.

Now, one of the aspects around managing pricing, inventory, non-inventory, labor cost and so forth, is the ability for businesses to understand the details of job costing and profitability. And it is like any other businesses profitability. We used to think in the past that it is more downstream than businesses look into their profit and loss, their accounting, their focus on profitability.

But over the last several years, we identified that businesses want to get visibility on profitability upstream as and when they're operating on a job as and when they're performing operations, they want to know, is this job going to be profitable or not profitable? And it is based on how much labor time, how much parts and materials are getting used, whether these are parts and materials from the inventory or a flat rate pricing. They want to know whether this job is going to be profitable.

If profitable, do they have a gross margin of over 60% or the gross margin is under 60%? And they, each businesses treat their jobs and tasks very differently depending on whether this is going to be profitable or not. And just to extend that on the profitability, it is not just based on the jobs.

They want to know which technicians are more profitable than others. If some technicians are more profitable, is this because they're more skilled than others, then which customers are profitable? Which regions offer more profitability than others?

And then which, um, annual maintenance or subscription plans are more profitable than others. So profitability is an integral part of these businesses and they're looking into slicing and dicing this based on different pivots and understanding the details of profitability. So who in these organizations kind of wakes up and has this epiphany and thinks the lead on solving the problem using tech?

Is it a CIO, is it a business unit leader? Is it the CEO or who kinda has the insights to say, Hey, there's a better way of thinking about this? Yeah.

The person who's in charge of the operations usually is a VP of service or a CIO in the organization. But they are collaborating closely with CFOs of the organization because, you know, CFOs are still thinking that field service is my cost center. And the CIOs and VP of service are trying to prove this and transition this and transform the organization into a profit center.

So there is a, this constant kind of collaboration that I required. Some cases there is a healthy friction in some cases there is a better collaboration between these business units, but mostly it is the VP of service who is in charge of, um, these operations and ensuring that they're driving efficiency, reducing costs, and offering that great customer experience. Right.

So what's your ultimate best advice to organizations about how to go about doing all this? 'cause I think on the one hand, they'd all intuitively get it on the other side of it, they're probably a little overwhelmed and there's so many AI things these days, it's hard to focus on how to even think about maybe just getting the ball rolling. So when do you see people doing well today?

Yeah, so, you know, my advice to digital leaders, whether it's a VP of service, a CIO or whoever is in charge of operations in a service organization is threefold. The number one, which is most important is embracing agility. The pace of change is only accelerating and it's mind boggling.

The kind of changes that's happening in the industry, what seemed like current and cutting edge is becoming outdated very quickly. So as a business leader, it is extremely important that they are very quick to adapt, experiment with new technologies, and take an iterative approach on their process. And the most critical aspect is not to be afraid to fail, but they need to fail fast, learn and evolve.

So embracing agility is the number one advice to the business leaders in service organizations or otherwise. Second, which is again, super critical, is to ensure that the businesses are focusing on the why, because a lot of businesses are jumping onto adopting technology and tools without really asking the question of why do we need to adapt to a new technology? So always starting with understanding the core problem that they're trying to solve and really keep their end customers as defining the value chain for their business, the operations and the team and the why should always drive the technology investments in the companies and adopting to the right technology.

And the last, but not the least, is to prioritize the human element. Even with all these advancements of automation and ai, people are going to remain at the heart of their operations, whether it's field technicians on the field, back office users. And it is important that the businesses are investing in training their teams, empowering their teams with the best tools and fostering a culture of continuous improvement.

I strongly believe that technology should augment human capabilities and not replace them, and it is empowering the workforce to be more efficient and engaged. But these are kind of three advisors that I'd give to, um, any leaders, especially in service business. Just to recap, number one is embracing agility.

Um, prepare to be adapt quickly, experiment with new technologies, failing fast and learn and evolve, focus on why, and then prioritize the human element. All right, folks, you heard in here, hey, there's an old proverb about living in interesting times. And the thing of it is we're already there, but the other thing is innovation right now.

This is as slow as it's ever gonna get, so you gotta adjust starting now. Hey, no, thanks for being on the show. Thank you so much.

It's amazing. And thank you all for watching the latest episode of the Textron AI Leadership series. You can find this episode and others on our website.

We invite you to check them all out. Until that, we'll see you next time. Hello everyone.

Thanks for joining us today on the inaugural Red Hat Cloud Fridays with a WX session where you get real talk about real solutions. This is going to be an introductory session where myself and my colleague on video here from AWS Thanos es eos. Did I nearly get it right?

Thanos? You can introduce yourself almost. You can just say Thanos, just say, okay.

So my, my colleague Thanos is here to join us to talk through what we are doing with the Cloud Fridays event and introduce you to why AWS and Red Hat value the partnership between the two organizations. So much the way this session's going to go, we're gonna talk about the event in entirety and the different sessions that you can enjoy today if you choose to. And then go on to the partnership itself.

Highlight some of Red Hat software, which is available through AWS, and how that partnership and that software can really help you as customers, giving you some real examples of case studies to give you a guide to what you could expect from that benefit. So the itinerary for today is as follows. You'll have this introductory session for about 30 minutes, then there'll be a break where an interactive session with a coffee maker will go through some ideas for how you might be able to get your caffeine fixed, uh, for the next events that come along during this crowd Friday.

Then afterwards, we're hoping you'll, you will gather with us for the first of our breakout sessions. The breakout sessions are designed to have two sessions running simultaneously. The first breakout sessions will be in one stream, a session about Red Hat OpenShift service on AWS, where we'll talk about accelerating and innovating to deliver value at speed using Rosa.

And then the second session that runs concurrently, we'll be about Red Hat Enterprise Linux, where we will, um, talk about beyond the standard unlocking re's full value on AWS. You can choose to go to either session and we will of course make the other session available if you are interested in both topics to catch up with later during the second breakout. You can see there that, uh, we've got a session, um, that's called your Fast Track to it Automation Genius, where we focus on Ansible automation platform and how you can consume it and best use it in an AWS environment.

And then at the same time, also we will have a session about maximizing your marketplace spend with AWS. As I said, you will have to choose one or the alert if you stay with the breakouts, but you will be able to catch up with the other one later as a recorded session. Then we are going to, uh, wrap up the event with a real time, uh, question and answer, um, session where we talk about takeaways, answer any questions that come up, and, um, then move into a very informal possibility to have a networking lounge.

The networking lounge, um, will be available for you voluntarily. And there if you want to come in and speak to us as presenters, um, in an informal way, talk about the way you are using software and how you might want to utilize it in different fashions, then we'll be available there to talk to you, um, face to face. So that's what we're doing today, but let's talk about the really essential, uh, partnership that Red Hat and AWS has formed and why we have the day today to talk to you.

Well, Thanos is the best person to talk to you about the power of AWS. So over to you Thanos. Thank you, Simon.

Thank you Aaron. Thank you for the introduction. Um, I'm es and I'm super excited to be joining, uh, this event, the Cloud Fridays from Red Hat.

I'm representing AWS I'm the partner development specialist for, uh, red Hat at AWS. And I'm here to provide you with a brief overview of the world's most comprehensive and broadly adopted cloud, and also talk about our partnership with Red Hat as we are jointly addressing customers, including some of the most, uh, the largest enterprises, fast growing startups and leading government agencies. AWS almost 19 years old has experienced a remarkable growth over the past several years.

Yet we estimate that only 15% of IT workloads have moved to the cloud, indicating an enormous potential ahead. Our mission is to help customers like you lower costs become more agile and innovate faster. Let me explain why AWS gives this transformation through five key differentiators.

First, it's all about functionality. AWS has significantly more services than any other cloud provider. Most, most importantly, we have the deepest functionality within those services.

Whether you are running basic, uh, web applications or complex AI workloads, we have the right tools optimized for both cost and uh, performance. This makes it faster, easier, and more cost effective to move your existing applications to the cloud and build nearly anything you can imagine. We have also been the world's largest, and I may add to the most dynamic cloud community with millions of active customers and over a hundred of thousand partners globally.

That gives us the right to claim experience with every type of customer and use case experience that is always augmented, uh, by a wide selection of system integrators. With deep cloud expertise and a much broader collection of third party software you can use on top of AWS, security remains our number one job, and this is a central pillar around which AWS infrastructure and services are designed and managed. AWS is architect to be the most flexible and secure cloud computing environment available built to satisfy the strictest security requirements for the military, global banks, and other high sensitivity organizations.

As we commonly say at AWS, we've built our services with secure by design principles from day one, and that means including features that are, that are setting the bar high for our customers default security posture. Of course, let's not forget our ecosystem of security partners and solutions through AWS marketplace, which enables customers to create a multi-layered defense with security technology and consulting TE services from familiar solution providers they already know and trust. Fourth, innovation, innovation is in our DNA.

We pioneer several as computing with Lambda, we democratized machine learning with Sage. Major AWS is innovating faster than anyone else, especially in the new areas such as artificial intelligence and machine learning, the internet of things and quantum computing. Our base of innovation is continuously accelerating, accelerating with our customer obsession to help customers solve problems faster, leverage the latest technologies to experiment and innovate more quickly, differentiate their experience and transform their business.

What is unique about our innovation approach approach is that 90% comes directly from customer feedback. We listen, we learn, and we build exactly what you need. Finally, we have experience with 18 years of operating at global scale, running a wide variety of use cases, allowing you the flexibility of choosing how and where you want us to run your workloads.

With bid infrastructure spanning 114 availability zones with 36 re within 36 regions, we process millions of requests per second, maintain consistent load latency, and have proven that customers can depend upon us for the most important applications now and to the future. And not to forget that we are still expanding. That's one of the most proven operational expertise out there at greater scaling of any cloud provider.

Our unmatched experience, maturity, reliability, security, and performance is something that we can depend upon for most of our applications. These are not just claims, they are backed up by numbers. And our customers, 90% of our fortune of Fortune 100 companies use AWS.

Whether you are looking to reduce costs, accelerate innovation, or transforming your business, AWS provides the most comprehensive, secure and proven platform to build your future own and a natural choice for many of our partners to deliver their offerings, solutions, services. Upon. At AWS, we say there is no compression algorithm for experience.

What does it mean? You can't shortcut the lessons learned from running infrastructure at a massive scale for every imaginable use case. This experience from AWS translate translates directly into reliability performance for customer workloads.

And with that word in mind experience, I'm coming into the partnership between Red Hat and a WSA strategic synergy that delivers a trusted enterprise grade platform for accelerating cloud adoption and modernization across hybrid environments. Leveraging Red Hat's exp expertise in open source solutions and AWS comprehensive cloud infrastructure. Customers can confidently run their mission critical workloads on AWS while maintaining enterprise support and security with support for both traditional and containerized applications.

Through Ray and Rosa plus integration with Red Hat products on AWS uh, marketplace, the partnership enables seamless workload mobility to AWS Red Hat and AWS can help your organization run smoothly on AWS cloud, migrate your VMs to a new platform, simplify hybrid cloud management, and adopt a comprehensive AI portfolio to support each, uh, each stage of the AI journey With Red Hat, we share a common vision and have a longstanding relationship with a partnership. Going back to 2008, a partnership that has recently deepened and strengthened focusing on hybrid cloud innovation, virtualization, and AI deployment. We have signed both parties have signed a strategic collaboration agreement to propel virtualization and AI innovation across the hybrid cloud supported by the new marketplace availability and jointly optimized cloud platforms and synergy between our companies that enables you and your organizations to navigate the complexities of digital transformation more efficiently, ensuring you remain agile and competitive in an increasingly technology driven landscape.

And with that, I will pass it back to Simon to explain more on how this, how Red Hat can help you more within this trip. Thank you. Thank you, Thanos.

That's a fantastic introduction to the power of AWS one party in this really powerful partnership. So give me a few moments to explain how powerful Red Hat is too and why a vendor like AWS would really want to partner with us. Well, of course, red Hat's very proud to say we are the world's leading open source IT solution provider.

Obviously a few years ago that was proven out by, uh, IBM's investment into us as an organization. And, um, at the point we were invested into, um, these metrics, um, really, uh, show how powerful Red Hat had become. So at the time we were a 3 billion open source company from dollar point of view.

And of course, over the years since that, um, purchase by IBM took place, we have only grown even greater, um, most organizations globally and particularly the large ones utilize our software in some description. Uh, we have nearly 20,000 employees now across most of the globe. So as a vendor, we're very proud of our position and what we can bring to the AWS partnership.

So Thanos has already told you how important AWS sees this partnership, but I always like to show this slide to, um, really call out, um, how powerful senior leaders in AWS understand this power ship, uh, this relationship to be. Um, the particularly strong part of this side is the quote down at the bottom from Andy Jassy, one of the leaders at um, AWS. And he calls out the fact that because Red Hat has such a global reach and has revolutionized, um, Linux services within organizations and open source capability, we are obviously an important vendor to them.

And you can see there with the metrics on the right hand side that over 60,000 organizations customers of Red Hat work on the AWS platform with us in the partnership. And the Thanos did mention we have been partners from very early on, A quick spot check might be to ask anyone if they know when AWS delivered its first services. So that was only, I believe, around 2006.

If Thanos nods, then I think I'm about correct. I almost 19 years, almost 19 years old almost now. Yes.

So within a couple of years, red Hat saw that their customers and the wider it market was seeing great value in working with AWS on their global service delivery. As you can see here, it's, we've got an example of how much our partnership has innovated since its inception. So over time we've added more and more of our capability, obviously concentrating initially on our Red Hat at Linux Platform RL, but then bringing in other services as they became more applicable to our customers.

In recent years, AWS has worked very closely with Red Hat to bring those services even closer together. So, um, in 2001, I actually, um, joined Red Hats shortly before the release of Red Hat OpenShift service on AWS, the Rosa service that Ana mentioned earlier. And that's a very special moment in time from a Red Hat point of view, because it was the first service that Red Hat and AWS delivered jointly to our customers to give them the value of Red Hat software, but also supported directly by AWS and Red Hat site reliability Engineering underneath.

Over time, we've enhanced that. Um, I've got a call out for the new, very new, in fact only just announced that reinvent in December, um, the new managed service of Ansible automation platform that's now available to customers where, um, using a W S's Cloud Red Hat's able to deliver, um, the managed control planes for customers Ansible, if the customers are finding it difficult to manage those environments themselves. And recently we have even expanded this relationship to allow distribution partners to help partner organizations deliver value to end customers when they're looking at purchasing with a value add partner.

Um, and those distributors are able to help us in that overall process of sale on AWS's marketplace. So that tells us why the partnership is important to the two organizations and who we are. But I'm sure you were asking yourself, what actual software products can Red Hat sell us via AWS?

And this Blue side is actually an AWS slide. I stole it from a deck that AWS presented, uh, last year at the Red Hat Summit Connect events where we go around important cities within our regions and, um, our important, um, partner vendors present their solutions to our customers. And this side points out that actually those two major forms of delivery of Red Hat software to customers via AWS.

There's, um, the original, as I like to think of it, capability where AWS from their console sell on behalf of Red Hat Two customers are software services baked into EC2 instances that they deliver on their cloud. Now you can see there that very heavily concentrates on Red Hat Enterprise Linux. And over time we've expanded the service to include different facets of Red Hat Enterprise iNOS to give customers the most powerful capability possible to utilize Red Hat software through AWS console.

On the right hand side, we can also now sell Red Hat software through AWS marketplace. Thanos did call out the fact that AWS have innovated throughout that nearly 19 years of their existence. And one of the fantastic innovations they've brought to market over the last 12 years is their e-commerce marketplace solution.

This drives, um, customers towards the ability to buy independent software vendors, such as Red Hats software packages to package on top of AWS's, um, networked global services. And in this case, red Hat has now been able to, over the last couple of years, roll out more and more of our software to become available via AWS marketplace. It not only includes the, uh, red Hat solutions that are available on console, but it also includes OpenShift services, Ansible based services, some of our middleware services such as JBoss and, um, even extensions such as being able to take, um, a third party, uh, Linux maybe based on CentOS, an open source project, which was based on Red Hat Enterprise nx.

But now that those services no longer get a level of support, we even offer a listing on the marketplace that allows organizations to migrate, um, those CentOS workloads into rail on the AWS marketplace. I'll also call, call out a recent innovation where Red Hat has actually embedded some, um, predictive AI capability, such as a large language model in the IBM Granite Model and, um, projects from the open source community such as Instruct Labs, which allow organizations to build prototype test and deploy their first predictive AI workloads to their customers. Um, and that comes in the form of Red Hat Enterprise in Ex Server ai, and that is one of the listings that's available on AWS today.

You can even buy non-software products there as well, such as Red Hat Learning subscriptions and our support packages, sorry, consultancy packages very quickly. Um, it's very difficult to explain to organizations the difference between AWS console consumption of Red Hat software and AWS marketplace software. So I tried to build a couple of slides to give you an example of how the two things work differently, because sometimes you may have a choice of buying the same solution such as Red Hat Enterprise zenex from either capability in the first case, the case number one, let's concentrate on the AWS console offerings.

That's actually Red Hat software sold by A-W-S-A-W-S. Um, take that product against the defined price and deliver it to customers directly. They sell the product to the customers and they provide support for that software directly to their customers with Red Hat acting as a third line backup support agent on AWS's behalf.

If you then move to concentrate on how the marketplace works, it is subtly different. The seller of record for a marketplace listing is actually Red Hat themselves in amea. It would normally be listed as Red Hat Limited in this case because it's Red Hat that's selling you the solution.

Even though we are using AWS market, red Hat has a responsibility of delivering all the support for that software solution to the customer, and importantly, to many organizations, red Hat is able to take into account different pricing models for a customer in the form of what's called a private offer, which will recognize customers' commitments to overall usage of Red Chat's software over time, rather than being a standardized pricing through the initial console or PayGo offer, um, that we discussed earlier. Importantly though, red Hat's trying to make sure that all our software's available to you. So here's a quick table which shows you that you might be able to buy some of these products from the first option or the second option, or actually you've got a choice between either.

But importantly, all of them are available today to you via AWS. I'd like to take a moment though to call out my fantastic new service that I'm very proud of. So I did touch on it when I was describing the large list of software that's available from Red Hat via the marketplace today, but I really want to give a moment to concentrate on that technology.

It's a bit of, bit of a pet project for me. So this solution was released, as I said, announced in December at Reinvent in the us. And what this solution is, is it's run by, uh, red Hat and we're integrating AWS's, um, services to deliver the control plane of Ansible automation platform straight to our customers.

It's billed via the AWS marketplace. Private offers are available for organizations if they want to enter into commitment, um, contracts with us, and importantly, any commitment they make is recognized against, um, PPA private purchase agreements they may already have with AWS as well as that agreement with Red Hat. And that starts to dig into the customer value that AWS and Red Hat are bringing to you in our partnership.

Essentially through the partnership, we want to be able to deliver a reach to our customers, the benefits of AWS and their powerful, um, cloud Global services cannot be denied. And by being able to deliver Red Hat's technology combined with that capability further extends the ability for customers to enter into an open hybrid cloud approach. Many customers today are moving to the cloud, but many of them feel they still need to deliver assets in more traditional models, maybe on premises.

And when you combine the capability of being able to deliver Red Hat software in both those locations in the same way through the same procurement tensions, we are providing reach for organizations to migrate and augment their capability as rapidly as possible. And that's the speed that we think we bring to organizations. The managed services I've touched on, Rosa managed a a p the capability to have a managed advanced cluster security platform, wrapping your Kubernetes resources.

All these things are built with speed in mind and as speed to value for customers. What we are trying to do with AWS is squeeze the amount of time it takes for organizations to build their services and accelerate organizations into the situation where they're delivering powerful solutions on top of those services rather than concentrating on engineering the building of them. But at the same time, we are delivering all the control, visibility, and risk management that Red Hat provides with AWS.

We have great visibility tools that we will go into in more detail during the breakout sessions during the Cloud Friday, which help organizations make sure that even though they're traveling at speed, they're not incurring great risk. And finally, here's some of those case studies that I was talking about at the start. Red Hat's business on AWS is now massive, and we have some of the biggest organizations in the globe working with us in our partnership in Europe today.

Some great examples across all the different software products and services that we are delivering. And as you can see here from a Red Hat OpenShift service on AWS point of view, bp the well-known energy company, is delivering very powerful business transformation using our ROSA service in their organization. And we are very proud to say has created a full case study and video to, uh, walk customers through that, um, journey that they took.

And if you are interested in that story, the breakout session concentrating on ELL will give you much more details. The second example there, as I said to you earlier, I'm very proud of the managed Ansible Automation platform service, and it's just been released. And part of why I'm so proud of it is I work to help the Department of Work and Pensions in the uk.

As most of you probably know, the UK government is rather large, and the DWP is a very large part of that large government. And they have found delivering automation across their organization and all the, um, advances that brings to them as an IT operation, they have found it very hard to deliver across their whole infrastructure. And the managed ible automation service from AWS has given them the opportunity to centralize that delivery and rollout to their whole organization in standardized fashion.

Again, there is a session in the second breakout where myself and a great colleague from Italy discuss the power of this service and Ansible overall. And we will dig into that case study in more detail there. From a Red Hat Enterprise Linux point of view, we also, um, are very happy to say have another very large UK government body who is the largest pego consumer or console rail consumer for Red Hat, um, on AWS globally.

They use it extensively and they find the power of the dynamic capability to turn on rail services on EC2 dynamically whenever they need them, wherever they need them, um, a great value for their organization. And we will be, uh, driving a separate, um, session to talk about, um, the rail services as well during the breakouts. So with that said, we are coming to wrap up this introduction session.

I'll quickly remind you about the itinerary today and also take opportunity to thank Thanos for joining us. It's very, um, exciting to have AWS to come and speak to our customers at any point in time, and I think it really helped show customers the value of our partnership together. So what's gonna happen after this?

It's, thank you. So let's remind ourselves it's gonna be a short, um, breakout kind of session where, um, a coffee expert, we'll talk to you about your caffeine fix. You probably need it after listening to Thanos and I for this long in the introduction.

And then there's going to be the breakout sessions. Importantly, you can, um, during the session being delivered today in Cloud Fridays, you will have to pick one of the sessions in each of the breakouts because each pairing run concurrently. But we do encourage you to come back and watch the other breakout session as well as it will be made available as a recorded resource to you.

But those, it's important, Simon, I think it's important for everybody, anybody to access all sessions, all sessions. Thanks. Thank you.

And importantly, though, just, um, to underline each one of those sessions covers the different pillars of our software, which is Red Hat's, OpenShift, red Hat's, enterprise ux, red Hat's, uh, automation platform Ansible, and our capability to deliver value through the AWS marketplace. Then we will have a, um, a wrap up session where we will, um, give, uh, our feedback on the overall day, give opportunity to answer q and a if questions have been asked during the sessions, and allow you to ask questions directly of our expert panel. And then afterwards, we are going to provide what's called a networking lounge, where you can join those experts informally for a short while and talk to them about your, um, requirements, talk to them about your software usage or just come in to say hello and tell us who you are.

It would be wonderful if you could join us for all these sessions. And then, um, just want to leave you with some resources as well, if you follow these QR codes. There's a lot of information there about the ride wider, um, partnership that helps support this introduction session, um, separately from the breakouts, which will carry on for the rest of the event.

And that just sees us with a formal thank you. Thank you to everyone for joining us, and please stick with us for the great sessions. We have to come.

Goodbye. Enjoy, enjoy. Okay, good morning everybody.

It's very nice to be with you. Uh, my name is Matt Knight. Um, I'm the CISO at OpenAI.

Um, I joined OpenAI about five years ago. Um, and, uh, I've been building the security program there since then. It's great to be back here at RSA day one.

It's morning of Hope. We're feeling good long week ahead. Um, but if you caught my talk here last year, you may recall that I left you with with two points.

The first is that language models are tools that can help security teams. Um, security teams face many challenges in their work constraints, a lot of toil, um, in their objectives to protect, protect organizations and language models represent new tools and capability that can equip teams to move faster, be more effective in their work with less toil and drudgery. The second point was to buckle up, right, that the pace of progress in AI is, is is very fast blistering.

And as a result, we should all, as a community be ready for disruption and be ready to, uh, uh, to update in the pace of it. So if you've been following trends in AI for the past year, and it's hard not to, it's everywhere. I you sensed that this is, this is true.

I imagine we're seeing how LLMs transform how teams operate, and the the pace of, uh, progress has really only increased as we've, um, over the last year. And I wanna use this talk to talk about that, um, the, the progress that we're seeing in ai, um, and, and, uh, how we measure some of the capabilities, um, within it. So before we get into that, I wanna start just with a note on progress.

So I mentioned that I joined OpenAI in 2020, been there for about five years. Um, I was the company's first security hire, um, brought into, uh, to build the team. Uh, I joined right around the time that we launched our API service, our A as AI as a service, um, uh, uh, inference, API, and with it, a little known model called GPT-3.

Now, if you look at sort of the, the endpoints on that timeline, like g PT three in 2020, and now, you know, oh three in 2025, it's almost incomprehensible how we got from there to here, right? I mean the, the, these tools are so, so different from one another. Um, oh three is a model of broad applicability can be used for coding, for data, data analysis, productivity tasks, language tests like copywriting has applicability across, across industries.

Education, healthcare, finance, cybersecurity list goes on. Uh, it's multimodal capabilities. It really makes GPT-3 in hindsight in comparison look like a science project, right?

It's just these, these tools are so different. However, if you look at that, the, if you look at the timeline, right, the, the events on the timeline between GPT-3 and oh three, the points on the curve, the, you see that the progress all sort of leads from one to the next, right? So starting with GPT-3 and then reinforcement learning with human feedback, uh, codex, which was our first, uh, you know, coding model that we, um, and not, not the new Codex, CLI, the Codex we released in 2021, um, Dolly training, GPT-4, our low key research preview chat, GPT releasing GPT-4 four oh SOA oh one, the reasoning model paradigm, um, image gen oh three, um, all these innovations built on each other and collectively helped us expand and unlock new capability.

This is also true for, for cybersecurity capabilities. And this is something I get really excited about as a security practitioner. If we compare cyber capability eval performance across model families, we've observed that performance doubles roughly every 10 months.

Uh, we expect that this trend is going to going to continue. Um, and I'll talk about some of these evals in context later, but what I wanna start with is an anecdote in my time at OpenAI, I've gotten the witness and benefit from these tools evolving from these, you know, oddities, these, these kind of curiosities that, that, um, that, that captured our, our interest in imagination back in 2020 to real tools that offer utility, um, to teams, to teams like mine and, and, um, and, and industries, um, beyond cybersecurity as well. So I used GPT-3 pretty extensively in my early days at Open ai, but really primarily for, you know, basic like language tasks like, uh, you know, copywriting and, you know, in, in summarizing things.

And that's just because it, at the time, anecdotally, it just wasn't that useful for security. Um, but my, my true like aha moment with language models, um, came, uh, in summer 2022, and that was when we were training GPT-4. Um, so, you know, if you've worked with, uh, you know, ML or ai, you know, so the, the, the way that you train a model is you take, um, so your algorithmic knowledge in the form of source code, you take your training data, and then you, you, you run this training process over large amounts of compute, and it takes time, right?

And you start from, uh, a model that has like little capability and over the training run, the capabilities saturate. Um, so they increase over time. So my team got our hands on a partially trained snapshot of GPT-4, um, and, you know, we wanted to experiment with it and see what it could do.

So, so this wasn't the final form. This wasn't it fully, fully trained. It hadn't been post trained or, you know, had RLHF or some, you know, the methods that we used to make the model more ergonomic or useful applied to it.

So this is very rough, very raw, and we knew that it wasn't as good as it was gonna get. So we, um, we got our hands on on this model and we wanted to experiment with it and put it through its paces and just see what, what can this model do for us? 5?

Um, are there new frontiers that we can expand into? And there were two experiments that we ran that, that, that, that really did it for me. The first was we got our hands on, um, on a certain data set that had made, its made its way online.

So this was 2022, um, earlier that year, I think it was earlier that year. Um, there was a threat actor, this group called Conti, this ransomware group, um, that had been, um, disrupted. And as part of that, their, their internal chat logs wound up online.

Um, you know, this is like, uh, you know, these, these, these threat actors, these operators talking to each other about, you know, what they're doing and, you know, targets they're going after. And, uh, you know, sort of how, how they're operating, um, trying to, you know, do crime and, and, uh, take advantage of people. And, you know, it's a big data set, right?

Just imagine, you know, like big chat log. I forget if it was IRC or what protocol it was. Um, but we got our hands on this data set and we ran it through GPT-4, and we were able to ask questions, uh, questions like, you know, what, uh, what, what companies is this group going after?

Um, you know, what, um, what techniques are they using? Um, who are the people in the group and what are the relationships to one another? If I wanna defend against this group, what should I, what IOCs or or types of tray crash should I look out for?

And GPT-4 did a pretty convincing job, a pretty good job of surfacing real actionable information to us from this data set, pulling these needles from the haystack. And what was especially interesting about this is that this, um, this dataset wasn't in English, it was in Russian. And, and it wasn't just in Russian, it was in like Russian internet slang that these like operators were, uh, were using to talk to each other.

Um, and, you know, this was a, a real update for us that, that this was a tool that was gonna have, have real utility for us as a small security team at the time, um, and, and help us be more effective in our work. The second experiment we ran is a little bit more, um, applied. Um, and that was, um, or a little bit more, um, more technical, and that was experimenting with using GPT-4 to analyze, um, commands and, and, uh, in security logs.

So, you know, security teams, you know, just sort of as you know, we spend a lot of time, uh, a lot of our attention building systems, um, and methodology methodologies to, uh, analyze security logs and find signs of, um, of intrusion or abuse be, uh, whether it's, um, you know, in, in real time to detect or, um, you know, in the context of an incident to put together the, the trail of what happened. Um, and it's a, it's a massive, you know, data analytics problem. Um, so we're, we're, of course, we were curious, can we use, can we use our models to help ourselves be more effective in this domain?

You can imagine taking like a bash bash history or like, uh, you know, all the interactive SSH um, uh, sessions that happen over the course of your organization and using a model to, to analyze them, that would be super powerful. Um, you know, analysts, um, you know, time is really valuable. Um, they themselves miss things.

Um, you know, asking somebody to read, you know, all the, all the bash history, um, all the security logs in, in a company is like, just that would be cruel and unusual. Um, so can we use language models to help them be more effective, right? And, um, here, here are just two examples.

So, um, here in, in small font is an excerpt from a bash, you know, just a, a, a bash transcript of a system administrator setting up a web server, something that, you know, is pretty, pretty standard. Um, here's the same, uh, command, but with a little bit of added value that, um, uh, command in bold is a reverse shell in and pearl. And, you know, we threw a whole bunch of examples at, at this, um, at this early version of GPT-4 and asked whether, uh, whether there was, um, suspicious activity in the, I forget what the exact exact prompt was, but was on the order of is there suspicious activity in here and does it merit alerting the security team?

And again, in 2022, we found it did a pretty convincing job of, of, of, um, of, of, of, of telling us that there was potential here, right? So these, these two moments together, right? Um, were a real update for me.

Like these, these showed that the, that this technology was just beginning to be, was beginning to become some something that, that we as a security program could really lean on. And since then, it's really just taken off, um, as more model families have, have come out and as capabilities have improved. So two has the utility, um, in the security domain, um, and we're just getting started.

I, I sincerely believe that we're in the first inning collectively of, um, of the development of these tools and what they're gonna do for us as practitioners. Um, so with that, I want to, uh, transition into the, the, the next section here, right? So those are some powerful anecdotes, um, but, uh, but they're just the beginning, right?

So use cases and anecdotes are great, but how can we measure and evaluate these capabilities, um, uh, and, and, and understand them on a more, more fundamental level? So I'm gonna share a bit about how we do that at Open ai. And, um, I, I wanna start just by saying that significant credit for this work, um, goes to my colleagues president and former, um, who are dualhead between the OpenAI or who, who were dual haded between the OpenAI security and preparedness teams, um, specifically Joel Parrish, Andy Applebaum, and Olivia Watkins.

They're just incredible researchers and scientists. Extraordinary, we're lucky to have them. Um, so I need to, need to lead with that.

So first, let's think about how we test humans for skills, right? Um, the, the way in which we test humans is actually pretty narrow. Um, think about like the SAT for example.

Um, in the majority of it's a multiple choice test, um, which is really just testing for, you know, factual recall. Um, if we look at empirical and experimental, um, tests like, uh, you know, lab exams in high school or college, um, you know, those two, um, are are somewhat narrow as well. You know, you might, you know, do the, the standard high school physics lab, or you attach a, attach a mass to a piece of ticker tape and drop it and try to reverse out the, um, you know, the force of gravity or whatever it's you're testing for.

Um, you know, that test, you know, on its own does not directly extrapolate to, you know, other forms of science, but the method, the method does. And, you know, we can do that with humans because humans are generally quite good at generalizing and can fill in the gaps. It's just, you know, how we learn.

It's how, how we work. Now, if we look at, you know, sort of conventional methods for evaluating language models, um, we find that there are limitations with, with these methods. Um, so for example, some of the first evals for, um, um, for, for, um, security abilities were, were pretty narrow.

So the massive multitask language, understanding MMLU for computer security, um, was, was really just testing for factual recall, which, um, we know that models are generally pretty good at, um, but again, that's not necessarily a test that will generalize in a way that's super interesting. Likewise, when we look at, um, empirical testing methodologies, we need to consider the status quo of what existing tools can give us. Um, you know, things like, uh, pls L map met minica, like all these tools are out there, they've been out there for a long time.

They, they offer significant, um, capabilities, um, right on their own. Um, so this like, uh, web app exploitation example, it's just a, a really trivial, um, trivial code comprehension and SQL map has been able to do this since like 2015. So, um, you know what, what I would ask is what Alpha can language models provide over that?

Same too, if you look for like certain types of, you know, trivial spot, the bug examples like memory corruption, find your sources and syncs, you know, look for, um, memory allocation, um, you know, porn, arithmetic, things like that. Um, you know, it's reading comprehension arithmetic, um, and doesn't necessarily get you beneath that to the layers of complexity that one needs to take to generate, to generate like a, a real practical working exploit. Um, uh, which, which is, um, which is reproducible.

So our approach, um, this, and I just wanna first start by saying that this is all part of opening as preparedness framework. Um, if you wanna learn more about these methodologies, um, it's published online. And then if you go and look at the system cards, um, that we publish along with models, we, we go in depth in terms of how, how these, um, models, models perform in these different areas.

And, um, there, there's a lot of material we put out there on this, so if there's some interest, you should take a look. So, um, can I just briefly talk about two, um, types of tests that we were on? The first is, um, again, capture the flag activities, and the second is, is, um, some range testing that we've done.

Um, so we start by first taking as many CTFs as we can get our hands on, uh, and then down select to ones that are really interesting. Um, so, uh, today we have, um, uh, well over a hundred, um, CTF examples. I think it's several hundred examples at this point, um, in a very, in a variety of different, uh, uh, d different variety of different categories across the Mitre attack FRA framework, um, that we, um, that we, um, instrument models to, to go and try to solve.

And that gives us a, um, a reproducible, um, battery of tests that we can run, um, to evaluate capabilities as we go. So things like, um, web app exploitation, reverse engineering, um, different like crypto challenges, things like that. And with each of these, we're really looking for, for, for a few things.

Um, one is we want a working test environment, right? So that we can run this, these tests reproducibly. Um, and the second, um, is we're looking for examples that require non-trivial exploitation, because again, we're looking for that, for that alpha, that, that, that extra, um, that the uplift that we're looking to measure.

Um, we've got, you know, a bunch of examples online, but, um, you know, here's what it looks like in practice. We've got, um, uh, oh one preview on the left, um, uh, going after a, um, reverse engineering challenge. Um, it's able to solve it oh four on the right is not, and we have, um, you know, partial coverage when looking at the Mitre attack framework today, and we're looking to increase this as we go.

So let's go back to that slide I shared earlier. Um, and, and I'll just, you know, provide a little bit more, um, insight into the indices. So, um, the yellow bar that you see is performance on, um, high school level CTF challenges.

Green is collegiate level, and blue is professional. Again, this is just our classification of how hard these problems are. We see that, you know, there's been steady progress across model families from GPT-4 to four oh to oh one, um, uh, across all the categories.

And if you look at them on a timeline, um, I believe it's doubling roughly over 10 months or so. And if we look at oh three, we see there's a significant step up, um, again, um, across these categories. Um, and this snapshot is taken from the system card, which if you want to go read about, you can go, you know, reference this chart and, um, and, um, uh, learn about the methodologies as well.

So, CTF puzzles are great, but they, they really heavily bias towards, um, exploitation style challenges. Um, and you know, on the Mitre attack framework, I think exploitation is just like two or three cells. It's a small, you know, small subset of the landscape.

So how can we test and identify other forms of trade craft that we're interested in, such as identity based attacks, um, you know, the ability to, you know, operations plan and move laterally towards an objective, things like that. And for that, we, um, are building out our own cyber range, um, and where we can put together a collection of systems and, and have scenarios that we think are a little bit more representative of, um, uh, than just, just what the narrow CTF count challenge can, um, uh, can capture. And you can read all about this in the O three system card.

We've, um, you know, published pretty extensively, um, our first, uh, couple scenarios and how we've, we've gone about constructing this. The initial results here show that there is a ton of room to go that, um, you know, the models really have not been able to, to put it together in a way that is super interesting just yet. Um, the model can succeed if helped.

That's what those collections on the right are. Um, that reflects the scenario where we, um, you know, give the model in its context, uh, and in the environment, um, basically tools that help it solve it. Um, but if you don't give it tools, you just give it hints or you give it nothing, um, it's not able to to solve the, the puzzle.

But if you wanna know more about this, I will direct you to the, um, the system card paper. It walks through a few scenarios the team has built, um, uh, how the model does on them, where it succeeds, where it fails. I think it's super interesting.

Um, and, uh, I want to emphasize that we're really just at the beginning here, and this is an evolving science. Um, it's one that, um, you know, our preparedness team, um, uh, you know, spends, you know, full time working on this, you know, so how do you, you know, sort of anticipate the next, uh, sort of, you know, battery of tests that's gonna be really interesting and meaningful and, um, and, and build out that methodology. And they're always looking, uh, for partners here.

So this is interested and you wanna get connected with them. Um, come find me after and I'd be happy to get your info. So I wanna just quickly talk about, um, you know, what these capabilities mean for us as defenders, right?

Um, so as you know, just as we saw from GPT-3, uh, to GPT-4, we, we sort of had new frontiers, um, open, um, op you know, be, become open to us as defenders. We've seen, uh, many more, um, along the way since then. So I just wanna share a couple, um, examples of how we use these tools within, within open ai.

So, um, access management is a challenge that every organization, um, encounters in some way. Um, whether it's access to documents or cloud resources, like your employees are gonna need access to stuff to do their jobs. And so many organizations wind up building services that enable, um, self-service, um, authorization, self-service, access to things, because without it, you wind up with human bottlenecks, you wind up with like an IT team in the mix of having to like, answer access to, uh, tickets, which is like slow, um, is not great for security in that you have sort of central, um, you know, humans without context making decisions about things that they maybe don't fully understand.

Um, so we built a service called Access Manager, um, which is a platform to provide scalable access management, um, across the organization, um, and what we built GPT-4 into it. So, um, so that if you're a user who's looking for a resource and you don't know, you know, maybe you don't know which, like, uh, um, I am group to request, or, you know, you know which resource to go look for. You can just go ask in natural language.

So maybe this is, you know, I'm looking for, you know, you know, this, this document, or you know, just as I showed at the top here, you run a command and you get an error message. Maybe you just copy and paste the error message into it. Um, the model is able to, you know, just with knowledge of the groups that exist, um, within the, the, the, um, the environment is able to suggest to the user groups that may help them find what they're looking for.

Now, what happens if the model gets something wrong in this case, um, is nothing because we still have humans in the loop, um, providing oversight over it. So the model is playing matchmaker. Um, but then the approval that's required is defined by policy.

Uh, the policies attached to the resource. So, um, you know, for, um, some like low sensitivity resources, the, um, you know, it's, it's a lower bar, but for some things that are more, um, that are more, um, you know, more sensitive, you, you'll have, um, you know, either, um, a group of approvers or a dedicated approver who still has to review that Quest request, say, is this appropriate? Approve it or not.

So, um, it gives us that like belt and suspenders oversight around the model. And I know that this example, like isn't very cyber, it's like not very technical. Um, it's just simple process automation, um, that can help.

But, but that's why I like it, right? Things like this can help organizations make, make better, faster, and more security impacting decisions. And if you like, integrate the small improvements that like something like this can, can do for, you know, maybe you're nudging a user towards like a lower privileged group or something, then the, the administrative role that they know is gonna get them access to it, but is gonna get them access to a lot of other stuff.

You compound that over time and that can add up to meaningful risk reduction. As a security leader, I wanna empower my engineers, um, and analysts to focus on the, like, the highest order of work. The most important tasks is they can be, they can be doing at any moment.

Um, and we all know that security engineers and security teams have to do a ton of legwork before they can make security impacting decisions, right? Um, this is especially true for, um, detection engineers and incident responders. Um, so we can use language models to automate parts of this process, um, to help our teams, you know, be more effective, be in more places, and focus on the bits of work that we need them to do.

So we built some lightweight automation to go out and interact with employees and, uh, and gather information, um, during investigations. So suppose an employee takes an action within the organization that introduces some sort of an insecure configuration, right? Maybe they, um, you know, they, they, um, touch an IM resource that is, um, that, that, that is sensitive or they, um, you know, share a document publicly or something like that.

Um, historically a security engineer would like, reach out to them and like, you know, try to engage, like, Hey, did you do this? Was this intentional? Um, that's a lot to ask of a security engineer who's probably managing many of these, right?

Who needs to go and sort of juggle this, this caseload going, you know, you know, make small talk, hi, I'm so and so from the security team. Did you do blah, blah, blah, right? Um, it's just a level of cognitive load on that engineer that, that, you know, you really want focused on, on stopping the bad guys and, you know, not making small talk.

Um, so we built a, um, uh, a chat bot, um, that we can point at, um, certain problems and use to gather information. So, um, here's what such an interaction looks like. So suppose a, um, an employee shares a document, publicly trivial example, we can point the point that bot at the person say, Hey, I'm the friendly security bot.

Looks like you did this. Um, have it go back and forth with them. And then we have the security engineer come in at the end, review the transcript, and, and, um, and then actually make the corrective action if they need to.

And what we found is that in many cases, the simple nudge from the bot is all that's needed to, to get them, to get the employee to say, oh, shoot, yeah, I actually, I picked the wrong group. Let me fix that. So in many cases, we come and we, we, we do that check and, and the conditions already remedied by the, the employee.

So again, it's like, this isn't like super technical or interesting. Um, it's, it's, I think, I think it's super practical though because, um, it helps teams be more effective in, in, in ways where they're, they're constrained today, limited time. So I'm gonna jump ahead to something that I think we're, um, to, to an area that is a little bit more technical, where we haven't really seen these tools, um, you know, fully come into their own yet.

But where I think the potential is huge. Um, and that is in, um, static analysis, finding and fixing vulnerabilities in source code. Um, it's a huge opportunity for LLMs, but it's one that, that, that really has not been, been realized, um, yet.

There have been some pretty interesting proofs of concept. But, um, you know, to, to my knowledge, I, I'm not aware of there being any like, super serious public sort of, IM, I impacts or cases of language models being used to improve security today. But I, I wanna talk about the, the opportunity.

So there are many static analysis tools available in the market today. Um, none of them are perfect. They all kind of, um, you know, have their pros and cons, but one thing that they, they generally fall short on is, um, their ability to perform across business logic.

You know, you might be able to, you know, they might be able to run a regular expression across, across the code base and say, look, oh, that looks like a log for J. You know, uh, uh, it looks, this looks like it might be vulnerable to like, log for J, for example. But, um, what I'd like, what I, I used to run an AppSec team, and one of the things that I always was testing for is, you know, is this information that I wanna put in front of a developer actually relevant to their decision making?

Is it actually gonna help them, uh, write better code? Or is this something that I, as a security engineer, is relevant to me? So, uh, language models, um, I believe have potential here because of their ability to reason and understand the context in which code is being, being written and, and, and, and deployed.

So, um, you know, if we think, you know, further ahead and, you know, a language model won't just have to look at like a single line or even a function or even the code itself. It might be able to ingest all of the, um, context about the environment that it's, that's that the code's being written within, um, where it's being deployed, um, your developer logs and documentation, um, your gire, you name it, things like that, um, that I think are gonna just unlock all sorts of new frontiers for what we can do with these tools. Um, but I wanna emphasize that we're just, just the beginning here, and we really have not seen this, this come into, into its own.

Um, it's an area that we're active actively exploring at open ai, and I know many others across the industry are too. And I think this is gonna gonna really be something one day. We're also partnering with, um, you know, with, uh, um, we're not just doing, doing it ourselves.

We're partnering with industry on this. And, uh, we're proud to be partners of DARPA's, um, AI Cyber Challenge, um, where we're helping to support, uh, the teams that are competing in this, um, with building, um, what they call cyber reasoning systems. Um, these, um, agents that can analyze code for vulnerabilities and, and fix them and patch them.

Uh, the semi-finals were at DEFCON this past year. The finals are are coming up this summer. Um, it's, it's, it's really promising.

It's really exciting and I, I can't wait to see how they, how they do this fall in the summer. So, um, just to, just to wrap up, I wanna just end by, by thanking my team at OpenAI. Um, I'm very lucky to be able to associate with these folks.

Again, special shout out to the OpenAI preparedness team and, uh, my colleagues, uh, Joel, Andy and Olivia from there, um, also we're hiring, so if, if you're interested or know anybody who's looking, we're really hiring across the entire company. Um, and, uh, if you have any questions for me, I'll hang out in the hallway for a few minutes afterward and would love to say hello. So, uh, thanks so much.

The six five, somebody is back. We are in our sixth year, and unsurprisingly, we're talking about ai. And this year it's really about making AI real for enterprises real ROI and real payback.

Daniel, how you doing, my friend? It is good to be here. You know, six years ago when we started this thing, we did talk about other stuff.

Um, it changed a lot in the last couple of years. I think it, like was maybe even the first year, there wasn't even a track for ai. By the second year, there was a track.

By the third year it was like a little bit of a theme. And about, I don't know, the chat GPT moment came and it just became the summit. And it's like, you go to companies, they're like, Hey, we want you to talk about applications.

They're like, can we talk about ai? You're like, well, you don't even have ai. It's just become one of those things.

But you know, that washing is kind of over now. And to your point, it's all about getting real and people actually getting value from these investments and, and are we pulling the hype path? Yeah.

And one company who not only is using AI internally, but also making AI real for enterprises is Box. And I'd like to introduce Aaron Levy, uh, co-founder and CEO of Box. Welcome to the show, Aaron, first timer, we really appreciate you coming on.

Thanks for, uh, having me. Good to, uh, good to be here. Yeah.

Seeing you on, uh, you know, analyst calls and, uh, watching you, uh, on other podcasts. It's been, uh, it's been a, you like to mix it up? I like that.

That's good. Um, it's, uh, you know, te Tech, uh, is moving quite, quite quickly. Lots going on.

So there's always, there's always some drama to, uh, to discuss. Exactly. It's a funny, funny, uh, little walk down memory lane, Aaron.

'cause I, I, I listened to you recently on with the other besties. You know, pat and I have probably each other's besties we're hundreds of episodes further along than all in. Oh, wow.

So, uh, yeah, I won't say they copied us, but maybe they copied. Um, but the, the, the, you know, we're manifesting maybe being like top 10 since Jason always tries to manifest being number one or whatever it's, he does. But Pat and I actually did make the maiden pilgrimage to Miami, and we went to the first all in Summit.

Oh, wow. So before it was cool. We actually went down there and You're the, you're the hardcore OGs.

We are, but I haven't been since, to be fair. We didn't, we didn't go back. But in the beginning it was very cool.

It's been very cool, and we've met these guys over the time. But you were great. Really enjoyed hearing you on that show.

Glad you were able to come over here. You know, we, we get a little bit more practical on the tech stuff here. We wanna talk about, you know, your company a bit more.

Um, so let's just kind of start off with this whole AI first enterprise. And, you know, we're hearing a lot about this. We're hearing companies coming out and saying, we're going all in on ai.

We hear other companies coming in saying, we're getting rid of all our people. We're just gonna be ai. Um, you seem to have a vision for what an AI first enterprise is.

Talk a little bit about what that is and the tipping point that sort of took you down this path. Yeah, so, um, so I, I actually, I think you laid out a, a, a pretty good continuum there. Um, uh, you know, there, there are some companies that are, you know, probably fully resisting kind of heads down, uh, you know, heads in the sand from a, from what the impact of AI is going to be.

And then on the other end, there are companies saying, okay, we're gonna only use AI and, and, you know, we're, we're gonna, we're gonna do our best to just basically automate out everything. And we're, we're, we're probably, you know, squarely in the middle. Um, o obviously leaning more toward the, the acceleration toward the future side, but with a deep belief that AI is really a technology that augments people, um, and will, will help us, uh, just have way more capability, um, both as individuals and as organizations.

And so we, we laid out our AI first principles, um, and it, it's been kind of the, the past year that we've, we've framed them and then recently we've, we've been more public about it. Um, and the, the, the core premise is we wanna use AI to do more as an organization. That means more innovation, uh, better features for customers being more responsive to customers, solving their problems faster, hopefully even proactively, um, being able to better sell to our customers so we know exactly the, the next set of capabilities that they likely would want, uh, be able to deliver better marketing campaigns.

Um, and those that are, are global, uh, by nature. And so when you think about that list of, of attributes, um, that that really is about doing more with ai, not doing the same amount that we already do at a lower cost. Um, and so I, I'd say as a line, a very clear line in the sand that I think most customers or companies will have to deal with, is, do you want to use, is your metric and KPI for success using AI to do your current operations at a lower cost or just vastly more as an organization, possibly at the same cost for even more because of the flywheel of you get AI productivity gains that get reinvested back into the business.

And so that, that's the core premise, um, of, uh, of our, of our principles. And then there's a bunch of kind of sub elements of, of that. So, um, you know, what types of platforms do we stand on, uh, standardize on as an organization, um, you know, in areas where we do automate some degree of work, let's say frontline customer support tickets, where we can more efficiently solve those with ai, how do we reinvest those savings back into the business, uh, to be able to drive better customer success outcomes?

And so our approach is really about how do we exploit AI to, to just do more as an organization and be even more competitive? Um, and, uh, and that that's our ultimate takeaway and what we're, what we're driving with, and you know, we're, again, we're seeing a variety of different options that that enterprises, uh, you know, around the, the, um, around the world are deploying. But our, our firm stance is use AI move as quickly as possible.

We want every single person in the organization to get trained up on ai. We want to have AI first, and, and sort of AI native ways of working across, you know, every job function. So this is even doing things like changing our, uh, you know, how we hire and our interview process and the kind of skillset and capabilities that we look for for employees.

But that's a little bit of a, you know, vignette on what, uh, what we're up to. No, I think it's important. I mean, listen, customer zero has, has been a, a real phenomena forever, especially when you're trying to sell other customers and convince them that a technology is the right one, uh, to adopt.

And, you know, as I've seen from, you know, mini computer to client server, client server, social, local, mobile cloud, right? PA lot of enterprises, they don't wanna be first. So at a minimum, uh, the customer, uh, the tech company who's, who's trying to propose this stuff has to be using it, uh, the themselves.

And I'm still dumbfounded. And, you know, as industry analysts we're like professional event attenders, we're, we're like circus clowns. We go from event to event.

And it still astonishes me when senior leadership gets up and they don't talk about the ROI that, that they're experiencing, uh, from their own ai. So yeah, it's, it, it's super important. 0 wiped out a a cadre of multiple companies, right?

And whether that's travel agencies, uh, classic stockbrokers, uh, big box retailers, and then we went to music and movie downloads, right? And that churn, uh, I think this generative AI and agentic, uh, is going to wipe out, uh, different companies as well. And I always like to say, you know, I get a lot of questions and it's gonna be the sea of cubes that are still there that wasn't transformed accounts of payable, accounts receivable, uh, and a lot of the, quite frankly, the backend work that really hadn't been touched.

'cause if you look at traditional backend SaaS, it's really been, I'm gonna sure a little bit of B-P-O-I-I, I get it. Okay. But let's take this, this process that was paper and fax, and we would shuttle it around on a cart, right?

And, and then we digitized it, digital transformation. Okay. But this one's, uh, uh, very different.

Um, yeah. And the, and the data, uh, has changed as well. I mean, for 50 years it's always been, uh, from a relational database, what can I stick into DB two or Oracle or, uh, an S-A-P-E-R-P system, a giant intelligence spreadsheet.

But really the bulk of the information that we haven't lit up yet, uh, is, are the documents, uh, are the emails, uh, videos and even, you know, customer service call recordings and, and even, even even videos. Talk to me about why this is such a, an asset, uh, and what can people do in this new, uh, AI era. Yeah, so, so I, you just hit the nail in the head.

So, so if you think about, you know, there, there's, there's, obviously this is, this is quite, uh, simplistic, but, but let's just say at a minimum, there's two major data types in the enterprise. There's structured data that goes into a database. You, you just kind of enumerated, you know, a bunch of those data types.

So if I wanna look up HR information, it's in a database. If I wanna look up my customers, you know, um, uh, you, you know, the, the amount of a, of, of pipeline we have for a customer that's in a CRM system, um, if I want to look up, you know, certain, you know, ID records of, uh, of our customer base that's in some database, so all structured data, queryable, synthesizable, calculable, you know, that, that, that, that's structured data. We've had, you know, since, uh, since the, um, you know, relational database world, uh, uh, you know, years.

So that data, uh, as you just kind of noted while it's, you know, absolutely some of the most important data in the enterprise, it's only about 10% of our corporate data. And we, when we, when we analyze kind of how much information is in an organization, it's about 10% of the data, 90% of the data is all this unstructured stuff. And what's incredible is that if you look at any graph of the growth of unstructured versus structured, the unstructured data is just truly exploding.

Um, you know, exponentially, and the reasons are everything we're doing in a digital world is producing some degree of unstructured data as, as an output of that thing. Every video call we're on that you record massive amounts of unstructured data, every, you know, video that we, we take with our phones on a construction site or, um, you know, doing a, uh, uh, you know, uh, you know, any kind of maintenance call that's unstructured data. Every, every thread in a Slack channel is unstructured.

Every document that we collaborate on, every invoice, every cad file, every design asset, all unstructured. And so 90% of our data is unstructured. It's growing exponentially, but it's had one, uh, sort of less desirable property, which is that computers, when, when, when left to themselves, can't really do that much with this information because until a person cracks open the file and a person looks at the document and a person watches the video, and a person looks at the design, you know, image, we, we don't get that much value out of that information.

And so, uh, for, for box, you know, we've been in the business of helping customers govern that data, manage it, ensure they can collaborate securely around it, enable the permissions and access controls around that data. Uh, but, you know, the, the life of that content is, is actually fairly brief from a value creation standpoint. Once it's in our platform, it's, it's, you know, when you're looking at it, when you're sharing it, when you're collaborating around it, and then it sort of gets stored and, and maybe you look at that file again every couple years if you can kind of go and find it.

So that means that the vast majority of our data, the vast majority of our time, the the vast majority of the time is, is sort of not working for us. Um, and, and we can't ask it questions, and we can't glean insights from that data that we can repurpose in the future in a very natural and easy way. So for us, this is where the Gen I breakthrough kind of, kind of happened.

And, and we, we got instantly, you know, instant religion on this, uh, from, from a chat GBT standpoint, and actually maybe similar to your, you know, as you introed the conference lineage on, on this, um, we had a big, you know, AI effort about seven or eight years ago. And, and unfortunately it sort of petered out because what we saw was that every single AI use case we had, we had to have a different model. We had to have a, we, we had to do a different training run for a different data type, and it was just never gonna scale.

And so, so the large language model phenomenon actually ended up being this massive breakthrough because these LLMs with hundreds of billions of parameters can basically handle now any data type, an invoice, a contract, a, a document, a screenplay, a, uh, a memo, every type of data that an enterprise uses, we've basically trained these models to be able to work well with. And so it's a massive breakthrough because now for the first time ever, when you connect this enterprise content with to ai, you can start to use the AI as effectively a reasoning engine for having the computer interact with all of this information. So you can ask any type of question of this data.

You can use AI to read the documents or listen to the audio or watch the video to extract the structured metadata from that content. And then as a result of understanding what's inside the content, we can now automate any workflow. So these are the three big focus areas we have, which is talk to your data, extract the, the most important intelligence from it, and then automate any workflow around it.

And for us, that's, you know, quite literally a 10 to a hundred x increase in the number of use cases that we can go solve for. Because now when you put your content into box, it's the actually the moment where the value creation starts, as opposed to the tail end of the value creation on that data. And, and it's just a complete inversion of the value proposition that we can offer our, our customers.

Yeah, it's really interesting because, you know, we hear Jensen, uh, from Nvidia talk a lot about kind of this next era of compute architecture, right? That basically the GPU era is gonna completely reinvent the way infrastructure ex looks. And what you're kind of saying is another, is the interesting point is the old compute infrastructure really only worked well with structured data.

The apps only looked at data rows and tables, you know? Um, and now it can look at everything and it's being kind of reset, redesigned where all this data can be sort of instantaneously accessed. That's what, that's the problem we're trying to solve for, because to your point, I mean, I've heard the data point, 99% of enterprise data has not yet touched ai ai.

So while we've already scraped the entire world internet for consumer use cases in enterprise, we're actually still super early. But inside solutions like yours, inside tools, uh, and, and, and, and technology, like what Box does a lot of that sits there where if you could actually query it, you could ask it a question. And of course, in the future, it might be the way we ask an LLM something, it might be the way we walk around with our Ray Band glasses and just talk to the environment and be able to tap all that data and start to interact.

That's really exciting. The, the other thing that's super exciting is, is like, is what we're hearing about agents, you know, I liked your question about kind of, are you going a 10 x or one 10th? Meaning do you wanna do the same amount of work with one 10th the amount of expense?

Or do you want a 10 x and maybe the same amount of expense agents seem to be the way they work 24 7. They don't, you know, they, they, they don't have weekends. They don't ask for vacations.

They, um, but at the same time, they work kind of alongside humans, at least in the beginning. I think there's a very, um, interdependent, uh, relationship over time. I think it will be more autonomous.

I think we all see that happening. Um, but like, you have to be thinking about how it's transforming productivity models there, and how is it actually changing the way a business can derive value drive productivity? Talk a little bit about that, because you started alluding to it a little with Gen ai, but like now when agents come into play, how do the use cases evolve?

Uh, and what are you seeing your customer base in terms of their adoption and utilization of ai? Yeah, yeah. So, so, um, so if you, if you, I mean, we, we almost have to rewire our understanding of ai, which is amazing 'cause we're only two and a half years into, you know, literally this, this movement, this modern movement in the first place.

But, but it was a little bit of a, of a mistake at, at first with AI to think about it as, as this sort of chat bot interaction. Um, that was just the form factor that I think, you know, kind of illuminated the potential of what AI could finally do. And the right way to think about AI is now, you know, to, to your point on agents, uh, really any, any ability to have an AI system that can go off and do work for you.

And that work could be as simple as just answer a question for me. So today's kind of chat to BT experience, but it could be as advanced and as comprehensive as, you know, a an agent that goes off and does multiple months worth of work in a matter of minutes or, or hours across any number of systems and any amount of data. So if you just think about that, any data in your organization, any, any type of work using any system in the organization for any amount of time, uh, all of those sort of variables now become possible based on the, the type of age agentic experiences and architectures that, that we're delivering and that we're starting to see in the market.

And so, so what does that mean? I mean, that, that's a fundamental sort of transformation of, of enterprise productivity, because now as a, as an employer, as a worker, I can start to think about what type of work do I wanna farm out to an agent that, uh, that, that I, that is sort of discreet enough where I think I can review its output and I, I have some ability to kind of comprehend what it came up with, but, but something that would otherwise have taken me hours or days or weeks to go and execute. And so I'm, I'm getting that much compression of efficiency and time when, when, when that agent goes and does that work for me, and it's gonna, you know, take some time for every organization or every team or every, every sort of job function to start to understand what those, those use cases are.

But, you know, they're, they're becoming, they're becoming, you know, more and more obvious as as time goes on. So if you're in an organization and somebody says, let's go research that market opportunity, and if the answer is not to come back in an hour with that research as opposed to a week or a month, which is what the answer would've been two years ago, then that probably means you're not being AI first, right? If, if somebody says, let's go write a strategy document for, for a new product, and that takes, you know, more than a day, then, then probably you're not, you know, fully exploiting the, the, the power of agents.

If somebody says, Hey, I wanna update, you know, our SDK from, from one version to another version, 'cause there's a library change in, in reactor Python, and the answer is not that that will be done by the end of the day, and instead it's, you know, two or three weeks, then, then you're probably not using AI agents for coding. And so this, this is where over the coming, you know, 1, 2, 3, 5 years, this is the ripple that will happen throughout organizations. And the impact is that we will collectively be able to just get way more done.

And the very optimistic thing is that we will get to work on the things that are absolutely much more value creating and much more interesting to do, because it's just not that interesting at the end of the day to go update a Python, you know, version in some SDK like that. Like that is not, that is not the work that anybody decided to become a computer programmer to, to go and, and work on. And agents will be the, the ones executing those types of tasks in the future.

Rapid question, because I know Pat's got a better one, but I do wanna ask you, you just said something about nobody became a programmer. What is the Aaron Levy belief on all the juniors at Stanford right now that are in computer science and engineering that are kind of hearing the rumblings that there, there, there will be no, or very little or a lot less need for programmers in the future? I'm just curious what your quick take is on that.

I, I'm, I'm still very bullish on programming. Um, uh, I, I think it's, um, I think first of all, understood, the, the leverage you now get with, with AI is, is, is just enormous. And so, so if anything, it's probably the best time in history to be a, you know, engineer because of the impact that you can have where one person can now do the coding of, of two or five or 10 people.

Um, and, and you can work on the much more interesting parts of problems. Um, I think it really pays to have a, a deep understanding of, of the core principles of, of engineering. Um, I don't think that, um, I, I don't think any of that goes away or becomes less important in an era of ai.

Um, you know, there's, um, uh, uh, Martine Cado, uh, had a great sort of, you know, tweet about a month ago, which is, which is, as an engineer, you always wanna understand the layer of abstraction below you. Um, so if, if you're, if you're, you know, if, if you're building, um, uh, if you're building software in the cloud, you really do wanna understand the cloud infrastructure and what it's capable of, or, or you're just less, you're less effective as an engineer. And so even, even as a vibe coder, if you don't understand what's happening in that system, then you're gonna be less effective.

You're gonna be able to u you're gonna use the tool in, in a less impactful way. So I do think that, that what, what what's gonna happen is, um, you're gonna see that the barrier to entry as an engineer goes down so more people can participate. That's what's so powerful about vibe coating.

The, the sort of impact you can have as a, as an expert engineer goes up. And so, so the ceiling sort of, you know, rises in the process. Dylan Field kind of came up with this analogy, uh, from Figma so that the, the floor lowers, the ceiling raises, more people can participate.

And I think we're only scratching the surface of the amount of software that the world actually needs. If you think about the amount of software that we could have for life sciences and healthcare and biotech and, um, and, and industrials and, uh, education, uh, uh, just we are, we are actually in, in a, in a, in a, from a societal standpoint, relatively low on the amount of software that we have, given the amount of impact software could have on our daily lives and our, you know, societal and business experiences. So I'm, I'm still very bullish on engineering.

Um, and, you know, maybe that changes, maybe my answer will change in 10 years from now. Um, but, but I'm, I'm extremely optimistic on the function. Yeah.

You know, historically, I mean, I've heard a lot of tropes. I mean, virtualization was gonna kill, uh, data center growth. And what it happened is it increased it by five x in 10 years.

Yes. Yes. I, I was around for desktop publishing and everybody was saying that all the creatives are gonna die, right?

My gosh, Harvard Graphics and PowerPoint was gonna kill everybody. Okay. And, and, and it didn't.

And, and, you know, final example was iPhone was gonna put all photographers, uh, and creative, uh, process processors, uh, out of business. And it, and it just, it created more new jobs there, there were jobs that were eliminated, but then we had social media. And do you Remember, do you remember in, in, uh, in mid two thousands, uh, the, the case, uh, that it was going away because of cloud?

Do you remember this one? Yeah, Yeah. And oh, oh, and outsourcing Yes.

To India too. I look At the mainframe was supposed to die too for the last years. So, so the thing that went viral for, you know, as, as viral as anything could go in like 2008 or whatever the year was, was this idea that the cloud is gonna render it, uh, basically obsolete.

Because in a world of cloud, everything comes to you for free and you no longer have to manage anything and, and you're just, you're just consuming these services. So the role of the CIO is sort of obsolete. The role of it is obsolete.

And, and what actually happened was the exact reverse because in a world of, of infinite abundance, uh, uh, curation matters, integration matters. Making, making decisions on what systems do you kind of go with, and how do you architect those systems. The bar on that raises tremendously in a world where I only have five or 10 vendors to think about or choose from, it's, it's actually not that strategic.

My, my integration work. 'cause there's only one way to do it. In a world where I have a million options now all of a sudden my, my talent level for how do I deploy it and get use of it goes up and, and, and sort of necessitates actually having a very strong function in that, in that, uh, in that area.

So, so, you know, I I, I could see the exact same thing happening where in a world of AI agents being abundant and, and work is sort of done for us at a much lower cost per unit of, of work, the bar for curating that work goes up the bar for making really intelligent, high judgment decisions about which work to go with. If I can have an a AI agent go and deploy research across 20 fields at once, now all of a sudden it's actually really important that I figure out which decisions to go with that those agents came up with. And so, so, you know, creativity, judgment, uh, and, you know, kind of human agency, these things actually start to matter a ton, which is why I I, you know, I'm very bullish on, on, you know, humans as a result of ai.

Yeah. Just to close the loop on this kind of where Dan started on the question about, uh, uh, uh, coding. And historically what we've seen, I'm gonna give you an anecdotal, uh, an anecdotal guy, my son who's 22, is an entry level ai, uh, software engineer, and he's doing work not entry level.

He's doing it, you know, for fifth year, right? And he's using the heck out of these coding tools to help him, uh, help him do that. So it goes both ways.

And Aaron, to your point on how you're hiring people on, Hey, tell me about what AI tools, it even goes back to the prior days where you would even put a mastery of Word or PowerPoint or, or Excel, right? But now it's gonna be all these great tools out there. So Aaron, this has been a great conversation.

I I wanna round this out and talk a little bit holistically, uh, about enterprise, uh, productivity, right? Today we have roles. Yeah.

If you have a role, you're expected to do this set of work, you have certain KPIs, and this is the work you do, how does this change in, in this new AI era? Yeah, so, um, I, I do agree that the roles will evolve. So, um, uh, you know, we have, we, we have sort of over time, uh, established a high degree of, of specialization by necessity.

So very kind of Adam Smith, you know, kind of, we want to compartmentalize the work. So we get efficiency gains by that work being done over and over again. You know, kind of very, very kind of clear division of labor, you know, sort of theory there.

And AI is sort of letting us as expand to a few of our adjacent functions. So the, the product manager can prototype their idea, uh, as a, you know, effectively as a front end engineer, the designer can, you know, kind of quickly, you know, sort of write the spec of their, of their idea as, as a product manager would have. Uh, the marketer can kind of take on two to three or five different types of marketing roles because AI can sort of augment, you know, their, their, their skillset.

So I do think roles will begin to evolve. I think we'll see some degree of collapsing of maybe some of the, the hyper-specialization that we've evolved to. Um, but, but I, I see that as mostly an optimistic thing because, because you know, at, at, at, at some point you do want to do more at inside of an organization, you do want to have a greater impact.

You do not, you know, you know, you, you, you infrequently want to be just the person that you go to when you want that little button rounded you. You wanna be able to have more service area that you go and cover. So I think this is actually gonna be better for employees because they can actually get more done.

They can, they can touch more surface area. And then what it'll mean also is this really incredible thing happens where it's almost a leveling of, of access to resources. So, you know, if you look at the, the kind of, of, uh, you know, sort of job functions and level of specialization a 10,000 person company will have, versus a 10 person company, obviously the 10,000 person company is gonna have a role for every, every permutation of, of every specialization that exists.

And the 10 person company just, just can't do that. They can't afford it, which means on some dimensions, they, they can move faster, but on other dimensions, they're, they're missing out on the person that can go and translate their marketing to a new region. So that means they can't enter more markets.

The access to, to this one, you know, sort of, uh, uh, you know, specialized engineering skill that makes it so they can't build certain functionality. AI is sort of the great neutralizer of those gaps, because now all of a sudden that 10% company can instantly have, you know, a set of SDRs that, that can help them grow faster. They can instantly have a set of marketers that have a specialization in some marketing that, that, you know, kind of program that they want to get into.

And so what's gonna happen is, is actually, it's, it's, it's a fantastic time to be a 10 or 50 or a hundred or 500 person company because you now have access to resources that were never affordable to you and, and that were never available to you previously. So these are the kind of things that economists, you know, can never really quite capture. Because, because they're, they're thinking about, okay, AI takes, you know, sort of, let's say AI offers a 20% efficiency gain.

So that means 20% of the labor force is sort of impacted. And, and now we're gonna see this sort of, you know, the jobs impact of this. But guess what?

The Economist never does. The economist doesn't say, wait a second, all of a sudden, the millions of firms below 50 employees have resource access in, in a way that, that we've ne that is unprecedented, where they have a marketing team, a sales team, an engineering team, at a scale that was never possible before. Those firms will now grow at a, at a, at an accelerated rate, which guess what it means, they're gonna hire more people.

I, you know, that the, the what will happen is, is you're actually gonna see firms expand with, with humans, because AI makes parts of their functions more productive, that allows them to go and reinvest as a result of that. So you're just gonna ultimately see this just shift the labor market as a result. And our roles in the process will begin to evolve, um, and in many ways just expand the kind of things that we can go into.

Yeah, economists don't get it because they don't actually work at companies, right? And they work in universities. That's my snark.

Uh, but it's like, I was in product management in 1995, and as a product manager in 1995, you did product management, product marketing, program management, and sometimes regional marketing. Okay? Like, you were responsible for everything.

And now there's a freaking group for everything of product managers. They're engineer ex engineers that don't wanna do this anymore core value. It it like, it just, it just got enormous.

And I think what it gets down to is being able to, I love, by the way, everybody's gonna think they're an expert in everybody else's. Yeah. Oh, it's gonna cause lots of problems.

I already do this, I already do this today with, with vibe coating is I'll, I'll say, Hey, what do you think about this, uh, this new product idea? I just prototyped it last night. And, and, you know, so it's, it's, you know, the CEO e is sort of now playing the PM probably.

And, and I think it's gonna, you know, cause a lot of, uh, there there'll be some interesting waves that happen in organizations. So It's actually kind of interesting, Aaron, because the way you tell the story, it's sort of really gracefully, uh, splits the middle, meaning that there's this kind of philosophy. You hear d uh, Dario from, uh, anthropic, you've heard Sam Altman all talk about like the one person billion dollar unicorns in the future.

And you didn't exactly say this, but then you also kind of see these mega companies almost creating efficiencies. I call it para. They're kind of, you know, pruning to grow.

Meaning you see like the biggest companies in the world are actually cutting head count 'cause of ai. At least they're saying they're, um, but on one side you're saying there's gonna be an economic boom that's gonna come out of these kind of small companies. Maybe it's not one person, but your point is like a company that used to be 20 people that was really capped and even Yeah.

Well, if I, if I can just, if I can just, uh, sort of highlight, uh, just, just the thought experiment, um, uh, about why this is like, so intuitive. So Facebook, Facebook has, has sort of been rumored that they're gonna do AI for helping you optimize your ads right? In their, in, in Facebook.

So I'm a, I'm a three person startup. I, I, I have some novel, you know, widget that I, that I sell. Um, and I want to, I want to, you know, two years ago, I want to go sell it online.

I have to become a, an expert in online marketing. I have to, I have to figure out what is the right message at the right time to hit my audience. What regions, what parts of the market, what, what campaign, you know, assets should I go and create.

And so all of a sudden my innovation is capped out because I am not good at marketing. And so the growth of my organization is sort of inherently slower until I can pick up that skill, until I can figure out what, what, you know, who to market to. Let's say Facebook, AI comes in and they basically just say, tell us your product.

Tell us how much you wanna sell it for. We will find you a customer base. We'll find you an audience that wants this thing.

It, it's, it doesn't take that much kind of, you know, imagination to think about, well, when that AI agent is sort of running these campaigns, it's gonna test a thousand messages. It's gonna test a thousand creatives, it's gonna find the exact optimal message for the exact right part of the market for the exact regions that you're in. And, and so if you're bullish on ai, then all of a sudden you, you, it's very easy to underwrite that that business will grow faster because of ai.

And if that business grows faster, guess what? That three person company is gonna be hiring more people for sales, supply chain, you know, management, all of those other functions. But the, the revenue per headcount, which is historic, if you look across the timeline of history, yeah, we're still seeing companies exponentially grow the revenue per headcount.

The s are so much bigger. You know, we didn't have trillion dollar companies or multi-trillion dollar companies. If you were a hundred billion dollar company and not that many years ago you were massive.

Now it's like, oh, you might be in the top 200. So we've seen this happen. And what I'm saying is, so the productivity has created this massive, and of course you've seen the money supply increase, the inflation, all these other things has created such a bigger economy.

So the point of a company now getting to a hundred million of revenue, or a couple hundred million, which used to be big yeah, can happen a lot faster with, with fewer people. But it still creates more people because as the AI grows, you still need more people. So maybe it's not about big companies adding tons and tons of jobs, but lots of entrepreneurial kind of fast moving startups that are gonna create lots of opportunities, lots of jobs.

Let's end this on an optimistic note, because, you know, AI does come with its bits of doom. Pretty optimistic to me. No, I'm saying that's very No, that's super.

I, I want Pat to come in. I'm gonna choose that. I'm gonna lay myself off and I, I, I have been waiting for the bot so that we can do this thing without actually having to do this thing, and then We probably Drinking beer somewhere together.

Alright, Aaron, on the way out. Thanks so much for opening. Give us the quick sort of what's your big, uh, how does box evolve with ai and what are you sort of most excited about in terms of the future of enterprise?

Yeah, I mean, so for us, the, the, the reason we're so excited is, is just back to this point of, of think about how much data in your organization or information in your organization is unstructured. And, and the, the, the question we ask our our customers now is, is what, if you could ask any question of all of your information, what, what would, what would change about your business? And, um, if I could know every, every single aspect of every contract we've ever signed, would, would, you know, the next thing to sell to a customer?

Would you know that there's a clause in a contract that, that you think is risky? If you could ask any question of your entire product feedback from every customer interaction, every log that you've ever had, would you build a better feature for them next? Would you, would you, you know, be able to auto plan your product roadmap and strategy that then you could go review as people, if you could onboard any person in your company?

And they instantly had access to all of the expertise of that organization. So they don't have to go ask around over a six or eight or 12 week period to random interactions that they have, but, but instead, they can ask a question of the entire corpus of information in the organization to get smarter and be able to jump into their role even faster. What, what would, what would that do to our productivity?

You know, could we make better products? Could we launch better marketing campaigns? Could we discover new, new forms of, of, of life sciences?

Um, uh, you know, could we, could we make, you know, bigger blockbuster films? That's the, that's what the power of our data, uh, you know, is and, and, uh, and what we're building at Box. So we're, we're incredibly excited when you, when you bring AI to all of this unstructured data and information, um, what kinds of answers you can now get from that, uh, from that content.

Yeah, I, I think the future looks really brighter and I want to thank you so much. Very provocative, uh, very thoughtful conversation. Congratulations on all the progress, success over at Box.

Good to have you with us here at the six five Summit. Let's, uh, let's do it again sometime soon. Awesome.

Appreciate it guys. Take care. Thanks For joining us for this day three opener at the six five summit.

com slash summit. More coming up next. Welcome To the six five Summit AI Unleashed.

I'm Dave Nicholson and for this enterprise AI spotlight, I'm joined by Rory Richardson, director of next generation developer experience and Gen AI at Amazon Web Services, which we all know by the power acronym AWS. We'll be covering AWS's customer-centric approach to ag Agentic ai. Welcome, Rory.

Thanks For having me. It's good, it's good to meet you. Thanks for being here.

Um, let's dive right into this. AWS has been involved with AI for quite a while. Can you kind of walk us through, uh, the history of how AWS got to where AWS is now in ai?

You know, we started this journey well before the rise of generative ai. We started incorporating AI into the fabric of our services, um, several years ago. So that, uh, things like optimizations or, um, performance analysis, uh, we've been using AI tools in the fabric of AWS for quite a few years now.

The generative AI journey, though, is fairly new to just about everybody, which was, um, uh, two years ago I think, uh, we went GA with Code Whisper, which I think was our first service that natively integrated generative AI into the fabric of the service, uh, back in April of 23. Would you agree that we're sort of at the dawn of the age of Ag agent ai, and if so, how about if you give us your definition of agentic ai? Oh, I'm glad you said it was my definition.

Not AWS definition. This is a definition that keeps evolving, you know, as, uh, the characteristics of agents continue to take on more properties or different properties. Personally, I like to land on the definition of agent with the three A's, but last week I added a fourth A, so I'd like to think of agents that, uh, is, is different than an application because it can form tasks autonomously.

It can do things asynchronously, and it can do so with agency, like in a pride and prejudice sense, being able to make some decisions on its own and have agency. But then I started to think about the, the rise of the super agents, which is what I see is coming next. And so I added a fourth word atomic because now we're thinking about agents being more modular, because we're starting to see the agent to agent communication protocols standardizing.

Once that happens, we're gonna be able to build more and more complex things, just like with Legos, by combining different agents to work together. So when I'm thinking about like, what is the, the scale or the scope of any given agent, I'm biased in that I try to, I tend to think of it in terms of like a microservice architecture and create an atomic, uh, mindset or framework for the agents so that we can combine them in a gazillion different ways to form super agents, which the term super agents, it think of it like a, almost like a persona based way to get something done. Um, for instance, think about, um, an SREA site resiliency engineer, a site resiliency engineer has as a human a lot of different things that they can do.

And these would each be separate agents sort of underneath the covers. You know, one thing that we learned, uh, recently is maybe we shouldn't, you know, spend a lot of time trying to teach people, different agents to go to, you know, like, uh, if I teach you, if you wanna do a unit test, review that slash test, wouldn't it be cooler if you could just ask, Hey, can you form the unit test on this corpus of code? And so it doesn't, we don't have to adopt an artificial language or another abstraction layer, but we can really get to natural communication just like you were talking to a human.

So When you use the term super agent, you're talking about an aggregation of agents as opposed to implying that there's an agent that's, that is orchestrating the other agents? Oh, actually the latter. I think about it in terms of, uh, being a concierge, okay.

So like when you go to a concierge at a really nice hotel, they handle everything for you. I don't know if you've had this experience, but I've stayed at a really nice hotel in China, and this, this concierge was amazing. You would just say, Hey, I'd like to have dinner at this.

At a nice restaurant. They would pick out the restaurant, they would order the driver, they would give you a translated menu, like everything would automagically be handled based on the intent that they understood of the experience that you wanted to have. So being, having an abstraction to handle the orchestration, picking what agents are best for whatever task, and then aggregating that result, well, it's more human, it's more natural in going from intent to the thing that you wanna get done.

So Rory, as we move forward with generative AI and other AI tools, what does the world of work look like for knowledge workers and, and for developers in particular? So I think we saw two things that were really positive. Uh, one, we, we were able to see that people were able to get up to speed faster.

So, for example, when we rolled this out internally, what we, we were able to see that our most junior developers, the NS, were able to, uh, learn our inferences and libraries faster. They were able to perform unit tests faster, get their documentation in, and even detect security vulnerabilities and remediation. So, effectively what happened was that people were able to be more productive faster.

Now, if you've been doing something, you know, a certain way all of your life, you're probably pretty fast at it. But getting, you know, those, those new folks on board, it can be tough. The other thing that we learned was to not necessarily focus on the hardest, most humanistic aspects of our work, but to think about all the stuff that we don't wanna do anyway.

Like documentation is a great example. No developer ever wants to write documentation. So if you concentrate your application of generative AI on the mucky stuff, then what you we see is transformative.

People are more excited about their work. They're more original, they're more creative because they're not searching through documentation that you don't really wanna be doing anyway. So there's two aspects, um, have been transformative to watch, you know, getting developers, uh, mid-level or, you know, accelerated really quickly.

Um, and then giving them their, their time back, uh, to focus on the work that is most important to them. And most interesting. I I, I don't come from a coding background, and so I, like, I love the idea of, of taking my thoughts and having code generated based on my thoughts, but I also really like the idea of then being able to look at that code and understand what I'm looking at, right?

And so, so, you know, that's a very, that's a very real thing that is not going to go away. Yeah. I, I, I'm a database person, um, as well.

And, uh, I would say when we started using managed databases, like in my world, it's RDS, everything that's a relational database service. Uh, I was not popular with DBAs because 70% of what A DBA does is super du duper boring. And, and, and this is, this is coming from place of love as a former DBA, but backups, patches, failovers are not that differentiated.

Not really. They're not unique to any given business. And having that taken over by managed service meant that we were shifting what A DBA actually does.

I mean, these people went to school for information architecture and pushing the boundaries of insights and information, and they were doing backups. So what happened when we said, all right, well, 70% of you know, where you're spending your time can be done by this tool. Then we saw the rise of data scientists and we saw the costs of data scientists go down because there were more people available in the market that could have a perspective on information architecture that were previously unavailable, or doing something that wasn't necessarily driving a lot of value back into the organization that was very unique to that person and that organization.

I see the same exact things happening right now, in particular with developers, is it gives them more opportunity to innovate and to do the things that are their most human, uh, uh, in, in creating something new, something different, something something that solves a problem in a unique way. These are all very human things, but writing a unit test SNOO Bill, I mean, seriously, no, no, no. Developer on Earth gets up in the morning, goes, oh, I'm gonna write 20 unit tests today.

Yes. Yeah. People, you know, no one is excited about it.

People clinging to it if they think it's the only way they can earn a living. Right? And so, as long as they learn that that's not the case, um, we should, we should all be okay.

But what, what's, what's, what's on the, what's on the generative AI frontier, uh, from an AWS perspective? What's coming, what's secret stuff? No one else is listening.

Just you and me, Rory. Mm-hmm. Mm-hmm.

Yeah. What's, what's coming? I'm totally not gonna get fired.

Secret. NDA, Let's take something you said earlier, like, is code going away? Um, because it's something that's very personal to me.

'cause I have a 15-year-old, and I've been trying to teach this kid to code since he was four. I use code monkeys. You get the monkey to the banana and you learn python over the course of 250 lessons.

No, I'm can't make a scrapbook, but I know how to write code. So I was going with my strengths on this one, and I'm, and part of me is like, I can't get those 10 years back. I mean, the syntax of writing Python specific, a specific language, when I look at my kid by the time he hits the job market, is that gonna be necessary?

Because really Python is an abstraction layer from the intent of what you wanna create to ones and zeros. And what we have seen repeatedly with generative AI in production is it compresses the abstraction layers, compresses the space between intent to what you're trying to do. So the syntax of the abstraction layers becomes less significant.

So in the fullness of time, do I think the role of Python has fundamentally changed Absolutely. Abstraction. The, the role of abstraction layers has fundamentally changed.

I mean, just think about natural language. Like, my kid's never gonna learn what a Jaron is. He'll never understand a dangling partisan, because basically he has a tool now that does all of that stuff for him, and it does it really reliably.

And really well just create the abstraction layer of communication. It's like, oh, here's a really nerdy metaphor. Did you ever take differential equations?

Yes, I did. Oh, Did you fail it? 'cause we all failed at once.

No, I'm just kidding. We didn't all fail it. Um, I remember the day that they allowed me to use a T 65 on the test, right?

Right. And I mean, that was four pages. And by the way, but First to learn how to use first you have to learn how to use that tool.

Yeah, Yeah. Uh, fair. But you also had to understand the fundamentals of differential equations, the humanistic aspect of applying a methodology or a mechanism and doing it well.

Now, none of that went away. Differential equations still around. But the four pages of handwritten mistakes that I would make on the exam gone, that's kind of what writing with generative AI is like, it, it just eliminates the possibilities for mistakes with stuff that has repeatable patterns.

So when I think about the future, uh, and I'm thinking pretty far out. 'cause I, I like to build for my kids. I wanna create the world that I want my kids to live in.

I love how their, their very mindset with technology has fundamentally changed compared to me. Like when I have a problem, I'm thinking, oh, I should build an app for that, right? Like, I'm trying to make a, an appointment with my dentist.

Wouldn't it be great if I could just, you know, go to the portal and make the appointment? My kids don't think like that. They think, oh, I would just contact one of these super agents.

They would stand up an MCP server that would communicate with the agent at the dentist, and it would dynamically and ephemerally just get done. They do not think in applications, which opens the door for well rampant hyper-personalization. If you are able to have an ephemeral or unique experience that is superior to a static experience, why wouldn't you do that?

So I grew up with websites. You know, you build a website and a bunch of people will come to it, but, and the experience isn't that different from one person to another. I mean, we, we've pushed the boundaries of it with, you know, companies like Amazon that have very, um, targeted experiences.

Sure they're personalized, but we're talking about hyper-personalization where it already knows that I'm going on vacation. It already knows where I'm going on vacation. It already knows that it's a hot place and it has curated a capsule wardrobe in linen so that I'm not too hot.

And those are the suggestions from my concierge type experience. This, this trend towards hyper-personalization is going to affect not only your retail experience, but literally everything that we do moving forward, it's going to be far less static and standardized and consistent and far more idiosyncratic. I love it.

So what did we learn here? We learned that if you're Rory's child, not only did you have to eat your vegetables, but you also had to write code. This is, this is amazing.

Hopefully, Hopefully, hopefully the, hopefully this will all turn out well with your optimistic view of the future. And I agree. I, I completely agree with you.

I think that's, I think that's where we're headed. Rory, thanks for joining us for this Enterprise AI spotlight. com slash summit.

On behalf of six Five Media, I'm Dave Nicholson. Stay tuned for more great coverage.