Techstrong Gang – July 18, 2024

Hello everybody, and welcome to the latest edition of the Textron Gang Army host Mike Baard. Alan Schiel is, of course, still on vacation, but he'll be coming back any day now. In the meantime, we're gonna be talking about Gartner AI and Hype Cycles, and we're also then gonna talk about the future of the IT team.

'cause they're gonna be able to build all kinds of interesting applications in a way that's much easier. Then finally, Bonnie's gonna take us through how, uh, finops and sustainability are about to converge. You're watching Textron Gang.

All right, folks, welcome back. Let me introduce the gang for today. We have Paul Nash, who's joining us from, I believe, North Carolina still.

Is that true? That is very true. Yes, it is.

All right. And then after that, Paul is the, uh, practice lead for application development for tuum, by the way. And then we have our own John Schwartz, who kind of works for tuum and Tech strong gang, but he's out on the West Coast.

John, how you doing? I'm doing well. It's good to be here.

And then finally, we have Mark Hinkle, one of the regular gang members is returning. He's one of our resident AI experts. Mark, how you doing?

Welcome to the show. Great, Mike, thanks for having me. And finally, of course, we have Bonnie Schneider, our resident expert on all things green IT and climate change.

Good to see you as all. Good to see you, Mike. Thanks.

All right. Alright, folks, let's get started with this AI stuff and the Gartner Hype Cycle. I've never been a big fan, and I think everybody kind of knows that, but hey, we have to measure something along the way.

And one of the things that seemed to come out in this Gartner hype cycle is it felt to me, mark, that the more realization is setting in about how long it will take us to get the benefits of investments in ai. It seemed like last year everybody was kind of maybe engaged in a little irrational exuberance. And so, are we growing up here, or what's going on, mark?

Well, I think that's with every new tech and shiny object, everybody gets excited. And then the framework for Gartner's, all right, I don't know about their insight is the, uh, what is the, the trough of inflated expectations. But this, you know, ai, the foundational stuff is happening today is real.

But when you see the crazy amount of investment in startups and new products coming out, or they're not really products, they come out as betas and people are like, yes, this is, um, you know, OpenAI, soa, everybody got excited about it, but we're not out making movies yet with OpenAI, soa. I think it's, uh, you know, it's foreshadowing and it happened in the internet. We all thought that the world was gonna change overnight.

And, you know, 24 years later we were reaping the benefits. I don't think that's the same, um, length that we'll spend for ai, but, uh, it's moving a little faster, but it's still the same thing, Right? com bubble, You know, Are the bubbles kind of exploding faster?

It seems like we're going Yeah, they are through this cycle. You know, That's a great observation, Mike. Yeah, they are.

But they, everything's faster, right? And every successive bubble seems to expand faster. I was gonna say burst, but expands faster.

And I think even the Garner analyst acknowledges that. There's a quote here where he, he mentions that the CEOs in particular tend to be optimistic and bullish about any new tech type of technology or novel technology. And about 87% of the CEOs agree the benefits of AI to their business outweigh risks.

Yet at the same time, I think only 4% of the businesses they talk to use AI to produce goods and services. And most of those come the highest concentration come in tech. So they are the early adopters.

But you're right, it's gonna take a lot longer for this to seem its way through other industries And all welcome the show. Have we always Kind of overestimated the impact that it actually has on productivity? I mean, it's interesting that Garner's talking about there'll be a boost in productivity, but if I go back in time, those productivity numbers don't really move all that fast just 'cause we have invested in it.

So how should we kind of evaluate our investments in ai? What's a more context about what exactly I it does for productivity? You know, I mean, this is a, uh, thanks for having me, Mike, on the show.

This is, this is a really hot topic. And, and when I look at, uh, my research, uh, and, and, and kind of cross or layered across the, the, the hype cycle from Gartner, there's a lot of similarities, but there's also some differences. And one of the differences that I notice is, uh, AI is being used to address some of the skill gap issues also being addressed to, uh, to your point for productivity and to increase the delivery of applications.

What we found nine months ago was there was a, uh, 18% of respondents to our survey indicated that they're using AI in their production workloads. We, we reran that study recently, uh, and after nine months, that number jumped from 18% to 54% of AI being in production workloads. So the adoption of AI is coming, it's, it's coming fast.

Um, why is that? Well, what we're finding is there's a need to drive, um, more applications and more insights to these applications, uh, and create these, uh, these, these, these tools faster. And by utilizing ai, it allows, uh, to these organizations to have an accelerator to do the job that previously would've taken, uh, more man hours to do.

So, I believe that there's a desire to, uh, utilize these, these tools. I think that there's, this is an evolution and there's certainly a maturity curve that has to occur, Uh, occur Here as well. I agree with that maturity curve, because I think that, uh, one of the obstacles in implementing this is, is getting, when you're leaning into AI to be more productive, sometimes in the beginning it might make you feel like you're being less productive because you're getting over, okay, can I AI do this?

No, that doesn't look right, let's try something else. Or trying different, um, leaning into different models to get better solutions. So it, it's tempting to spend a lot of time on AI and then say, well, this isn't actually making me more productive.

So I think getting over that hump is important. I think there's usually a large gap between what the IT people see as a possibility and what normal humans can kind of pull together. But Mark, I wonder if we overemphasize these large language models and maybe the great innovation that's about to come is these more domain specific language models, the smaller language models that are maybe easier to deploy at the edge.

And the second wave of this thing may be more profound than the first wave. Yeah, yeah. I mean, I think when they came out, um, the old adage, the, when you have a hammer, everything looks like a nail applies.

And so everybody was saying, oh, I could do this and that. And I think where it's really good is that helping us sift through data, draw insights and take the massive amount of data that we have and, um, get through it more effectively. But where we're at today, there's lots of talk about agents and automation and other things, and it's just, there's a big gap between its ability to, um, sift through data or generate images versus actually, um, you know, exhibit some kind of intelligence and reasoning.

And that's the gap that we're gonna see for the foreseeable future. Oh, one of the things that strikes me too is that it's not just end users where there's a gap with adoption of new tech, but developers, it takes them a while to kind of wrap their heads around things and how to use things. And I'm sure there's tons of proof of concepts out there, but what's your sense, uh, how proficient are developers these days incorporating AI into their applications?

Yeah, that's a great question. When I think about, um, the incorporation of the new tech stack and the new tools to an, to accelerate the job, um, businesses are, there was two things. One, we're, we tend to be, and just as human nature tend to be creatures of habit, we use what we know.

We, we, we, we, we perform, um, better with what we know, but organizations are changing their KPIs in order to release code more rapidly. Actually, we're seeing in our research that 24% of organizations are trying to release code on an hourly basis, but yet only 8% are able to do so. And that's largely due to the heritage tech stack that they're using.

So with that said, um, developers are, are basically looking at these new tech stacks in, in order to adopt new skills in order to do their jobs more efficiently, more effectively. And that's largely driven from the, um, demands that are put on them. There's also increasingly been a shift, uh, moving from, uh, day two, day one to day zero.

So shift that shift left kind of mentality of not just on security, but across the board, uh, testing, qa, uh, anything in the CICD pipeline for the SDLC, uh, is impacted by that shift left mentality. So developers are looking for ways to automate and use the right tools in order to be more effective in order to have their jobs, uh, to, to do their jobs. We're, we're also seeing that, um, maintenance mode, uh, for a lot of these applications is still a, a good percentage.

It's still a third of the time that's spent, actually, I'm, I'm sorry. It takes two thirds of the time for most developers is spending in maintenance mode, and a third of their time is spent on innovation. And that seems to be, um, an area where developers are looking at how to grow their own skill and grow their, their own desire to do the job.

So focus on innovation requires understanding new tech stacks requires, um, basically delivering code faster and applications faster. Hey, Mike, can I mention some of the industries that I think actually are gonna be early adopters? I know in the report they mentioned manufacturing and healthcare, but we have a story coming out, which just came out today on Thursday about contact centers.

This is kind of the contrarian view where they are actually adopting this. And one of the companies that I know is very, very engaged is United. They see this opportunity even among their customer service agents who are open to the idea of using AI to kind of reduce the redundancy in their jobs and let them do other things.

I know this, we always hear this, and there is a threat to some of their, to men, maybe many of their jobs. But there's, there's, there's, there's this value that they see in this, and maybe they start off, maybe this one industry that we look at as being most endangered by AI actually might be one of the early adopters and pioneers that show others how to use it along with human beings at the same time. Mm-Hmm.

I'm gonna be a little cynical. Are there certain industries where the customer service is just so abysmal that AI couldn't make it any worse? So what they not Yeah.

Yeah. That's, that's, that's part of it, right? You eliminate, uh, you know, the easy solution, right?

Mm-Hmm. That, that sometimes, especially where you can't find in the loop, right? Yeah.

And especially customer service where it's all about people complaining. You'd rather have ai, right? Listen, more patient than a human.

Yeah. I mean, I'll digress for a minute, but yeah, I had a misadventure involving an airline. I won't beat them up too much about it, but it was pretty clear to me that the issue was that the people behind the counter just didn't know how to use the software that they were being presented with.

So this task that they were trying to do for me was just way more complex and couldn't actually switch a flight standing there, even though I'm watching other people get their ticket switched because the other person seemed to know the software better. Well, That's, yeah, that's, that's, I mean, that seems to be the, the running pattern. All these, all these studies and surveys show, the training always lags behind the adoption.

You know, they, they throw this in technology in front of someone and say, okay, use it is gonna make you more productive. But the, the amount of training that goes into this just never seems to meet their standards. So, John, what is the mood in the valley?

'cause we've talked about this Goldman Sachs report on a previous show, and I don't think you're on it, but, um, it was pointing out that, um, the investment cycle for benefiting from AI for a lot of these companies is gonna be more extended than maybe initially thought. And there is a lot of money floating around in the valley tied to these AI projects. Will they all kind of survive, or are we gonna see them all kind of die in the vine for a lack of revenue because it's just taking longer than they anticipate it?

com bomb, the dot bomb era, there are gonna be companies that are just gonna go away, wither away. There are those who are gonna be acquired, they're looking for an exit strategy. When they decide, when they realize that they don't have the revenue stream they thought they did, a lot of 'em are gonna be squeezed by the, the big tech companies.

That's unavoidable, I think. Um, so there's a bit of panic that we'll set in right now. There's this kind of gold rush mentality, but then stark reality hits, and that's when you start seeing a lot of movements.

And most of it's gonna be bad. Yeah. One of the things I regularly enjoy when I do go out to San Francisco is driving down 1 0 1 and looking at all the billboards of companies I never heard of, and I have no idea what they do, but, um, mark, half Of 'em are gonna be outta business on a year or so after they put up their billboards.

I mean, especially like, there were, there were exceptions like Twilio and some others in Salesforce, but for the most part, you're right, Mike, it was like a, a, a marker landmarks of, okay, that one's gone. This one has no chance. This one probably has a chance.

You know, you could almost bet on half of 'em being gone. Mark, is there a big giant roll up coming or what? com era, but I think the technologies will, um, keep going.

I mean, I'm talking to you on a Chrome browser plugin using JavaScript, and that's an artifact of Netscape, which went public in 1995 and had a a hundred percent, uh, runup on its IPO. And you know, if we're talking internationally, global crossing raised, I think billions of dollars to lay these cables across the country, and they got bought up by pennies on the dollar. So the artifacts or the technologies and the infrastructure is gonna, um, live on.

But you know, anybody that takes the kinds of ridiculous amounts of capital and thinks they're gonna provide a return on investment, you know, I, I think I'd rather rather buy lottery tickets. Paul. Um, yeah.

What do you tell people who are developers today about, um, ai? Should they go work for startups or should they be looking more towards the companies that are gonna implement this stuff? And maybe that's where the action's gonna be?

Well, I think I'll take a, a little bit different stance to the, to the conversation here. Uh, you know, I, I, I focus, uh, as we know and kind of pointed out, I focus on the developer and the DevOps and the, and the platform engineering teams. That's my persona that I go after and I talk to.

And all the companies and vendors that work in this space, um, are building solutions and pr, um, solving problems that are addressing, um, anything in the SDLC, in the software development life cycle. Now, the, there's a big difference between what we've learned over the years and, you know, from historical kind of, uh, areas where things kind of rise and fall. Um, I think there's a big difference now is, uh, there, you know, there, the, the market in this, in the tech space is very cyclical.

And what we find is this innovation is being, um, born out of necessity. Um, if you take a look at the CNCF landscape, for example, if you go to Cobe Con or any of the associated shows around that, there's a significant amount of point solution vendors. And, and I agree with, uh, the, the folks on this call saying that, you know, not all of those vendors are gonna survive, and that's fair.

Um, however I do see in this space is because of the open source ecosystem and because of the openness of this ecosystem, there's going to be, and there continues to, there is now hap it's happening now, and there will continue to grow as a consolidation of not just partnerships, but acquisitions. So we're gonna see this, uh, constricting and, and kind of ebb and flow of the tech stack where there's a need to build out those, uh, point solutions to help drive, uh, the inefficiencies in the CICD pipeline, uh, or as, as well as the SCLC. But we're also going to see some of these companies basically emerge as the front runners, and those are the ones that are going to drive the business forward.

So you'll see, uh, just to sum that up real quick, we're gonna see, um, kind of that bifurcated approach where a lot of these tech stacks will be, you know, if they're not already an open source, there will be, there'll be open source, and then those, the, the commercialized version of it will be then incorporated into a way to monetize for these consolidation of these companies together. So, um, to answer your question, Mike, you know, it really depends on the developer's needs and what they're looking to do for their own career path. I think it's a, it's very rewarding to fix a problem that you don't, that you know is out there, which goes back to my point of developers like to innovate.

They don't like to be in maintenance mode. So if they're, if they're starting out in the startup world and they're innovating, that's attractive to, to developers, that's the direction I would recommend. But if you're looking for some more stability and some more, uh, focus on, um, product that are in market, maybe the larger companies, the right, right, right way for you, I would just say proceed with a certain amount of care.

Mm-Hmm. And John, you know, this and maybe this is one of the dirty Before, before it's gonna, yep. And then one of the dirty little secrets of it and venture capitalism is the VC firms are kind of like movie studios, right?

There's probably about a dozen of 'em that really drive, and they all kind of magically seem to come up with the same investments in the same area at the same time. And then there's like 15 companies that are all trying to compete in this tiny little sliver of a spot. And then, like a lot of the movies that the producers put out there, a third of them never get made 'cause they just die on the vine.

And, um, and then the other third probably get, you know, rolled up in different movies where different studios get together and they create a movie. 'cause they were discovered they were all making the same movie. Can I, yeah.

Can I make an analogy as remember, remember back in the late sixties when Easy Writer came outs? Mm-Hmm. It hit, it was something so different, so unique that all the studios decided, we're gonna make a biker movie, we're gonna make a counterculture movie.

They all tried, didn't always work. It often it didn't work, and it ruined a lot of studios temporarily. And I think the same type of principle applies.

So I think you're dead on about this VCIT dynamic right Now. That's not necessarily a bad thing 'cause it takes the risk of investing in new technologies off of, say, a major company that they're hoping will buy that company. And I think most of these startups are never really designed to go public per se.

I mean, when it happens, great, but for the most part, it seems to be, um, shall we say, um, you know, built to flip is about the best way I can describe it. But Mark, am I overstating the case? There's so much I wanna say there, Mike.

Um, I mean, I think that, I just look at it like tulip mania of the 16 hundreds is, you know, the, the, when tulips became a worldwide thing, it, you know, everybody wanted them. They're spending the equivalent of a year's salary on tulip bulbs, not individually, but that's just sort of the analogy. And everybody went crazy about tulips.

And what happens now is I can go in and get a pack of tulips for, you know, five bucks at Lowe's, and then they're everywhere and they're not special. So, you know, people are emotional when they're with their investments and this whole venture capital model, those guys get paid their, their management fees, whether they make money or not. And, you know, I think it's, you know, a matter of incentives.

The incentives are to put that capital to work and the entrepreneur's incentives are to make money, but, um, you know, they're playing against this with a stack deck. 'cause you just never can get escape velocity at these, you know, a billion dollar round for AI training, like scale just raised, or, you know, 400 million here, half a billion there. It's, it's, it's crazy.

All right, folks, I'm gonna leave the, uh, AI irrational exuberance conversation here and we'll be back in a minute. Discover how Cloud native is becoming the new compute stack at Cloud Native. Now on July 25th, we will explore the transformative shift towards modern containerized and microservices based applications.

Learn my Kubernetes remains a dominant force in container orchestration. And how WebAssembly is emerging is a powerful compliment. Don't miss this opportunity to stay ahead in cloud native development.

Register now and claim your spot in the future of tech innovation. Alright, folks, and we're back and we're talking about AI again, but this time with a slightly different spin. Um, AWS has been shown a tool that would enable a traditional IT team that might have built an app in the past to, uh, create an app.

Uh, instead of using a low-code tool and a database, or if you really wanna reach back in time something like Lotus Notes to build an app for the HR department, or they used all these extensions and there are a lot of internal facing apps that IT teams built. And this one is low-code tool, natural language interface, type in what you want, and then we will just pop out that app at the other end of it. And I think what's happening here, maybe, and we'll see, but um, the Paul, I know you were at the event, but the role of the IT team internally might be changing because if we do give them these Gen AI tools and they can pop out a new app in, uh, I don't know, let's say days, um, does the whole relationship of what that IT team that actually does for the organization kind of change and evolve because maybe that application backlog starts to just kind of disappear.

Yeah, I think that's really great. Um, uh, kind of segue to the, the, the delivery of these new applications and how these applications are coming together. So Amazon did, um, allow, uh, or, you know, announced last, what was it last week or so, the ability to, um, uh, reproduce these code these applications much faster.

And, and by doing so, there's, there's risk to doing that, right? When you, when you bring these applications and bring 'em to market much faster, um, you have to do, uh, sounding of these applications, or grounding, I should say, of these applications, right, to make sure that you're, you're, um, the applications are delivering the results you're looking for. And when we, you know, the announcements that came out of AWS and, and, and a lot of the other vendors are starting to do this as well, and, and making that developer velocity, uh, much faster by, uh, introducing, um, uh, ways to, to create applications without having to have the full on developer experience.

Now there's kind of goods and bads to that. Um, you know, it's that it's that whole garbage and garbage out model where if you don't have the right, uh, prompts or the right things you want to ask for, you're going to create a, a less desirable application. However, if you create, um, you know, know what you're looking for as the end user of that application or as the business, the line of business for that application, and you're looking to do develop it, you may produce the right prompts.

But what I like about the approach that AWS did at their announcement, as well as some other vendors are doing, is they're, they're doing things like fine tuning of, of, um, you know, for, for the, uh, for the data that comes in, uh, embracing things like Amazon Q and stage makers that kind of help, that kind of a driving of the, the user friendly interface that allows for that quote, citizen developer or the line of business to develop their applications faster. Um, but also doing things like, uh, connecting tools like bedrock and knowledge bases together, uh, allows for the AI creation. Like, so Mike, you were talking about that natural language and creation of those applications using, uh, prompts and such.

That's a, that's really a, an, an attractiveness as well as, um, I mentioned grounding, and this is something that early on and when AI first came out and, and it's getting a lot better now, but when AI first came out, the, uh, the, uh, there was a lot of hallucinations, there was a lot of, uh, data quality issues. Uh, this is getting better. And, and having those data sources, whether it's a large language model or a small language model, uh, from the core to edge to cloud or wherever that data may reside, um, grounding is, is a big part of it.

Um, and that's, that's an important part. But the big focus, or one of the big focuses from the event that, uh, that AWS was driving is this new era of builders and using the Q apps or Amazon Q apps in order to do that, allows for IT organizations to do that. Again, going back to what I said previously in the segment, um, focus, having the developers focus on innovation and rather than focus on maintenance.

Um, but this is a double-edged sword because if you create a lot of applications, somebody has to make sure that those applications are, you know, uh, properly regulated and governance and, and, and compliance and following the business practices and guardrails is a way to do that, right? To make sure that you have, uh, a way to, uh, pro provide that. And guardrails, just to be clear, there's a product called guardrails, um, and there's also just general guardrails in your organization to do so.

So as long as, uh, it is overseeing the creation, delegating that work out to the lines of business seems to be the, uh, desirable approach. I don't know, mark, I'm trying to struggle with this in my mind these days because, and I'd love to get your input here. The question has always been, so we have professional developers and DevOps teams, and they've been building applications for folks using internal and external, if the IT team now can go build their own apps using, um, a AI tool.

And, uh, Paul also mentioned another tool that AWS has where, um, basically they're saying end users can also create their own applications and, you know, they can use, that application will get generated by a few prompts that they type in, and it will create a personal app for you, essentially. And then that application, um, will be easier to create than it would be for you to go find one in a repository and actually learn how to use it. So they're basically saying everybody and his brother will start creating their own apps, and they might be redundant or whatever, but they'll be highly personalized.

But be that as it may, what will be the role of the professional developers and DevOps teams if end users and IT folks are building more of applications themselves? And how does that all come together in your mind? Well, I mean, I, I think that that it, for anything important, there still needs to be some lay level of governance and rigor and analysis.

I don't think that these apps will get that great for a lot of your key enterprise apps, but you may have, you know, um, I think John earlier was talking about the productivity and skills gap. I think at that's where those apps are gonna fill in. But at the, you know, end of the day, if you're touching enterprise data, uh, you know, highly sensitive, highly really correlated to revenue, I still think there's gonna be central ai, a central app, and maybe AI will just help.

Um, you know, as a copilot, this copilot model is probably more what it'll look like in an enterprise software setting. All right, so Paul, let's come back to that because, um, again, maybe being overly cynical, but there's a lot of apps that we ask professional developers to go build, and they get on this list of things that they're supposed to work on, and frankly, they're kind of boring, tedious, and not really a high priority, so then they don't get built. So will this kind of rebalance the system in a way so that those applications that people need will get built, but, um, developers will have more time to focus on the stuff that really matters.

Is that a way to think about it? Uh, I do. I think that's the right way to, to think about it.

And then also just kind of to respond to the, to the last kind of comments, I think that when we look at the cadence of delivery, um, there is less patience to go slower, and there's actually a desire to go much faster. So business KPIs are pushing, uh, for developers, DevOps teams, et cetera, to make sure that that applications get created faster. Um, so if you look at it from that perspective, there's a lower barrier of entry when you introduce, um, tool sets that don't require the, the highly skilled, um, developer to create.

Now what what this does is it frees up the, um, true developers and, and, and organizations to innovate and create those high value applications that do require, uh, you know, impacts to financial systems, et cetera, as we just talked about. But, um, I think that there's the lower, I don't wanna call 'em lower value applications, but the, the applications that don't require as much innovation, um, that can be handed off to, uh, others that allows for that workload to be distributed faster. And, you know, I I, I think that when we look at it from that lens, um, this opens up new opportunities and skillsets for organizations.

Um, and it also increases the productivity and, and efficiency because you're not putting everything on the developer or the high valued developer, uh, that, that that's focused on, you know, that that innovation, right? So that's, that's where I think that there's a lot of shifts in, in the business. Um, and, and I think that's really where we, we, we as organizations are, are kind of looking at these things going, okay, if we're going to move faster, you can't just keep dumping everything on, on the development organizations to do stuff because they just don't have enough time cycles and day, right?

So if you're going to do that in order to create those applications, uh, you need to create those applications in a way that, um, it can, the, the resources can be span, uh, expanded out to, to different, uh, different people in the, in the organization. I think that also creates a better corporate culture, uh, within each organization if they're, if the burden isn't being placed entirely on the developers to do these, um, more tedious things that can be, uh, done elsewhere, I think that creates just a better culture for the organizations to communicate within each other in their departments and to kind of understand how to be more productive overall. Well, 100%.

And there's less finger pointing, right? Yeah. There's less fingers.

So if you, if the requests historically comes in from the line of business and create me an app that does X, y, Z, and then the app doesn't do exactly what the line of business is asking for, now the line of business is responsible for creating their own app, there's then it creates for a healthier culture, Definitely. All right, well, let me argue the opposite of that. And John, maybe you wanna weigh in here.

So let me get this straight. One of the mantras coming out of the valley is that, um, we're gonna build more software in the next two years than we built in the last decade. Okay?

I can see how that's gonna happen. Um, but John, I don't know, does that just mean there's just gonna be a lot more crappy software running around there? We're supposed to manage, I mean, the people, This kinda goes back, yeah, it's, yeah, it's nice, interesting parallel with all the companies that that that reared their heads.

com and also during ai, there are gonna be some things that are gonna be totally silly there. Some, some things that are just never gonna gain any traction. I mean, even some of the stuff that's done in with the best of intentions, for instance, uh, on Wednesday, Salesforce announced it's, uh, first fully autonomous AI agent for customer service.

On the surface, it sounds really interesting and perhaps it'll work really well, but the only thing that that kind of I see as a downside is one of the analysts who was briefed on this told me that they did a survey on AI and language this spring, and only 6% of workers believe AI would automatically after them, within a, without a human in the loop. And basically, most humans are not open to a fully autonomous AI system. So even the best of intentions from one of the best companies doesn't necessarily mean it's gonna gain traction immediately.

So yes, Mike, in a longwinded way, what I'm saying is there is going to be some great stuff, but there's gonna be a ton of crap as well as you as, as we've seen year after year after year. And that's just goes with the territory, right? There are a lot of people with a lot of ideas.

Sometimes they don't have enough funding. Sometimes the execution goes slightly awry. Most of these companies, you think, even back to Uber, they started off as something entirely different before they pivoted to success.

So it's a, it's a trial and error process, and something's gonna come along from someone we've never heard of that will break barriers, break through walls will be incredible, but we don't know yet. And, um, that's why this stuff is so interesting to me. Mark, what do you think?

Are we on the cusp of, I don't know, for what to call it, AI software, crap fest and what's going on? Software is always a crap fest, Right? A software guys put the bugs in there for job security.

Apologize, Mark. All right. But seriously, I think that when you have an accelerant to the development of software, um, like we do with GitHub copilot or, um, open Devon and some of these other projects, we're just gonna, a, it's easier to make software, the infrastructure to host that software from things like for sale to new things that are coming out from AWS.

We're just gonna have more. And when you have more, I don't think that there's any, uh, magic bullet so far to make that software better. It's just, um, more software faster is where we're at right now.

So more crap to your point, Mike. All right, wait, wait, there's more. Can I not build an AI model that will fix the crap created by the AI models and the humans and there'll be like this other loop of things where I can kind of create another layer of AI that will fix things?

Paul, what do you say? I think that there's, you know, uh, checks and balances that have to go across organizations. I think gun know, when I look at, Um, the mature, I, I go back to maturity of organizations.

There are different el you know, uh, organizations that have, uh, fully mature and have fully implemented, you know, AI solutions all the way to the right. And there's some that are all the way to the left that are just getting started and not sure what they need to do. But there're the ones in the middle are the ones that are some, like, there's some point that they're trying to drive through.

And then when I relate it back to, I just did a study, which is somewhat related, but did a study on observability, and the study came back and it said that the tools that are used for observability is most organizations using six to 15 tools to get those observability solutions in play. Okay? Well, they're building more tool, they're using more tools to, to do the observability within their ecosystem.

But when it comes to ai, they're doing a similar thing, right? They're using the different tools as all these different promises to deliver the best of the best type of thing, but they all have their weaknesses and their different use cases. And I think if you look at, for example, um, uh, Claude just, uh, announced the, uh, availability of, uh, an Android app that harmonizes across web and iOS.

That's great, but it is Claude great for every, uh, you know, use case. I would argue no. Is chat GPT good for every use case?

I would argue no. But if you want to create, um, a tool or you know, a certain, say you wanna create programming, you may wanna use one AI tool for one thing, and you if you wanna create content creation, and you may wanna use another because it has that advantage over, but based on the, the build, the learning and the LLM and the backend. But Mike, to your point, AI layered on top of AI layered on top of AI actionable insights is going to drive what the changes would be.

So if you kind of, uh, I think the AI on top of AI conversation, that's more of the complexity that's being introduced, and that's going to have to change over time because organizations are not going to want to have this complexity, and frankly, they're not gonna want to have these fragile systems that are built on top of one another that I'll just break. It's a house of cards. All right, I'm gonna leave the conversation there, but there will warn everybody we are about to experience that proverbial thing called too much of a good thing.

We'll be back in a minute. I'm Bonnie Schneider, sustainability contributor to the Techstrong Group. I'm excited to introduce you to a groundbreaking new initiative from Techstrong Research, the sustainability pulse meter.

The pulse meter offers valuable insights into how environmental responsibility factors into tech purchasing decisions for key players in the industry. Position your company as a leader in the industry and differentiate from your competitors with the sustainability pulse meter offered exclusively from Techstrong research. Welcome back to the Techstrong gang.

Well, we've talked a lot about different ways to have more green operations, and that's kind of referred to often as green ops. And then we talk about being more financially responsible to optimize costs known as spin ops. But how are these two processes and practices working together, and how are they being app in applications also being able to work together to optimize costs, let's say, and energy efficiency in the cloud?

I took a closer look at finops and Green Ops. Hi, I'm Bonnie Schneider with your Ecotech Analyst Insights Cloud management is evolving with finops and Green Ops emerging as key practices, but what are they and how do they relate to each other? Let's start with finops, which is an abbreviated term for financial operations, uniting finance, tech and business teams to optimize cloud spending and green ops.

The focus here is to reduce the environmental impact of cloud operations. In essence, finops helps organizations spend wisely on cloud services, while Green Ops ensures those services are as eco-friendly as possible. US data centers currently consume 2% of national power experts predict this will double by 2030.

finops and Green Ops are not only counterparts, they actually work in harmony. That's because cost-effective cloud practices often align with eco-friendly choices. It's a win-win for saving money and protecting the environment.

So how can organizations get on board? Start by tracking carbon output along with expenses. Consider migrating to greener cloud regions or adopting serverless architectures.

Companies adopting finops and Green Ops will likely outperform competitors in both cost efficiency and environmental stewardship. So optimizing these two practices together can be demonstrated in many ways. For example, one is data tiering, where we're automatically looking at data closer to, or that we're using more often is automated, so that's stored in a more energy efficient area.

But what's also interesting is implementing it into energy use and cost as well. I mean, one study found that if you optimize and you reach out to energy just at certain times a day versus other times a day for computing, you're gonna save a lot in energy use and also in costs. And Google, Microsoft and AWS are introducing processes that are, have finops and green ops working together in their practices and in their offerings.

So this is a growing trend of presenting to the team of, uh, how to be more energy efficient as well as, uh, being more green in their processes. On one level, I love this, it is overdue in high tech. I mean, finops, you should be measuring energy.

Energy is a, a cost issue. And then it comes out to carbon, and it all should be one kind of metric that I can track or set of metrics that are in the same dashboard. On the other side of this thing, um, ops, it's quote unquote relatively new concept.

And basically it's it and, uh, finance teams are getting together to set a, define a set of metrics for tracking cloud consumption, infrastructure costs, and, um, and we act like this is a new idea. And the fact that it's a new idea is terrifying because I mean, what the heck have we been doing for the last 10 years? Basically spending money on the cloud without anybody actually monitoring usage and cost or any of that stuff.

I mean, mark, are we really that much of a set of drunken sailors that we don't know what's going on? I do think that, especially in cloud services, um, we've gotten to a point where it's, since it's usage based and because there are so many different moving parts, it's really hard to draw a correlation between the delivery of service and where the cost is. A good example would be in like serverless applications.

0005 seems like it's a sense per per action seems really, really cheap until you try and realize that there's hundreds of thousands or millions of those events happening every day. So I think it's, I think it's an observability problem and then a mapping to your, you know, finance. But I think we're getting much better at it than we were five years ago.

Paul, who's at fault here, and I asked this question because we've allowed developers to provision infrastructure at will on their own all these years. We gave 'em these lovely infrastructure as code tools and they told us that they needed to go faster and they couldn't wait for centralized it. And now, you know, we got cost issues and we're killing the environment.

So how do we fix this? Yeah, well, um, I think that it's, uh, it's very clear that you have to have controls and governance in place when you're talking about, um, self-regulation. Now the days of submitting a help desk to get waiting three days for, for that change to occur occur, those are long since gone.

I mean, organizations are not gonna tolerate that DevOps teams won't be able to reach their KPIs and organizations just won't. It, it just won't work anymore if you do to go back to that model. But with that said, allowing self-service provisioning, um, can, is a good thing provided that you put the, uh, appropriate caps on top of the self-service provisioning.

So if a organization says, Hey, I, you know, a DevOps team needs to provision more resources, they're, they're not t typically the ones that are responsible for the budget. So they're gonna over-provision every time, right? They're going to say, I just need the resources to do the job.

And you know, honestly, rightfully so, they just need to get their job done and get it, get it done appropriately. There needs to be, um, uh, awareness that, that organizations are, are putting in place, whether it's, whether it's human capital or software stacks that allow, there's plenty of tools out there that do, um, uh, analysis of the resources and utilization of those resources in order to, uh, uh, uh, provision the appropriate resources for those, uh, ebbs and flows of the, of the, of the workflow that that occurs. That is something that needs to be, uh, explored and, um, understood prior to doing the, um, the architecture design, it should be in the forefront of the design versus an afterthought.

I don't mean to disillusion you, but, um, what he basically said is developers have no idea how much carbon is being generated For application. That's right. You know, that's a really good point.

And I think that that's one of the things that that has to be looked at, which is why the tools that are being introduced now are making it for developers to be able to do the carbon calculations at the same time that that is being recognized from what I can tell by the vendors. So, um, make it so it's simultaneous, make the, the reporting and the, the energy use completely clear and transparent as they go. So it is an evolving thing, and I agree that would be the first thing that the developer might say, mark, Can we accomplish this with behavior and incentives and maybe asking people to do the right thing?

Or do we need to like come down hard with a bunch of regulations that start with the word thou shalt? So you said can, and the answer is yes, but will, and I think you hit on the right thing, is when we start incentivizing people to, um, you know, look at how they build things and how it affects the carbon footprint. Sure.

But the other part of that is it also going to align with the business goals is, you know, is the short term thinking of companies that are building apps and addressing, you know, opportunities in the market consistent with the long-term concern of like making sure that we don't turn ourselves into a sci-fi, dystopian, you know, bullet snow train or whatever, snow piercer or one of those kind of situations where we killed the planet. I think that there's, um, definitely incentives, but one of the things that also has been, comes into play is customers and employees within this company, within each company is are also driving this. So you have the green ops and the, and the finops coming together for those reasons as well.

So if being more sustainable is making the employees and the customers happy as well, and it's doing better, as you said for the planet, but it's also optimizing costs, that that's when it becomes a win-win situation. Right. John?

Oh, I was gonna say, Bonnie brought something up that was really interesting about the, the employees kind of driving some of these initiatives. I think that's really interesting and it's also really true out here, especially a lot of people choose to work at companies based on not just the salary and the brand name and opportunities, but basically also what the company stands for. And a lot of people over the years have shifted allegiance from one company to another based on what that company does.

If they're more green aware, which is long overdue, or they think about, uh, politically what they do on the side in terms of efficiency. I know it's like, Mike, I'm quite cynical about a lot of this stuff, but it is at least heartening it should have taken place years ago. But actually the fact that we are in this kind of finops era is, is encouraging.

I mean, before it was just wanting of use overuse of, of energy without any thinking of the consequences really. Um, so anyway, I just don't want to throw that in. All right.

Let me come back to, uh, wanted overuse of energy without considering the consequences. Okay. Um, mark, we're building more data centers than ever and a lot of this is being driven by ai Mm-Hmm.

And I can't help but wonder if, um, a lot of the compute resources being used to drive those AI applications warrant the amount of energy that's being created and consumed, because let's be honest, you know, did I really, really absolutely need to generate more carbon so I could write a better email? Yeah, I mean, well I think it goes back to our earlier point is that it's hard to, to measure and what Bonnie is surfacing there is vendors are now, you know, putting that correlation 'cause you're not consuming something that, that is easily correlated to how much energy it's using you, you know, you're an IT staffer and a big company and you have to bring in the next application and without, you know, connecting the dots. But, um, you know, right now, if you look at the sector that is doing the best around, um, uh, AI beyond technology as utilities, 'cause we're just sucking power and, and all of these, uh, data centers are, you know, maxing out very, very quickly.

Is that part of the problem here is that it seems like it's been an issue for all aspects of climate change, but we don't seem to be able to correlate outcomes to cause and no one seems to know for sure or can say definitively 'cause other people seem to argue about it left, right and center. But is this just part of a larger conversation where we're just not able to kind of connect those dots to Mark's point in a way that everybody will go, oh yeah, there's an absolute fact. Well, that's so funny you mention it because I think, I think that AI is actually one of the factors that's bringing this to the forefront.

And I just wanna also share this with the panel. If we're looking at AI and we're talking about, uh, how, okay, we know we're using more energy and we are building more data centers, so how are we gonna keep these data centers cooler? Well, uh oh, that requires a lot of water.

So that's bringing it to, and we can all relate to, to wanting to save water resources. I think that's universal. So that's one, one area.

And then it says, okay, well we know we need to keep these data centers cooler. Maybe we should build them in places where there's cooler just geographically like Finland. So that's what Google's doing.

So I think that because of the obvious ai, uh, demands on energy, the obvious, uh, need admitted use of water to cool data, it's, that's helping to bring this to the forefront where it's almost maybe years ago, well, we don't really know how much energy we're using, you know, like I think it was, it was, uh, plausible deniability perhaps many years ago. You weren't, yeah, you weren't having this conversation when during the cloud era, which it just kind of goes back to this, I'm sorry to interrupt, but Bonnie No, it goes back to this AI phenomenon where people are so invested and alert and learning about this technology and its repercussions and its impact that they think about other things like the impact on the climate, et cetera. Um, that's, that's something that didn't happen, I don't think as much, nearly as much with previous cycles of, of, uh, of, of technology waves.

Mark, follow me here for a minute, but I, it's awesome there's gonna be a data center in some cold place, but there's this thing called latency that gets involved and suddenly I gotta, I, I wanna run my AI model at the point where the data is being created and consumed. So I push it out to the network edge and that other Finland data center is just too far away. So now I gotta move closer to something and, and geography plays a, an issue here with the laws of physics, and then that messes up our whole energy equation.

So, um, do we need a a a better sense of the math here? Yeah, I mean, I think the, the thing I'll latch onto there is, is that whole latency thing is I think we're gonna see more AI at the edge. I think that our, um, personal devices will be able to handle a lot of the queries that we're sending to, um, data centers, and they're gonna run on, you know, efficient Apple AI chips and the chips that are from Qualcomm that are going into, um, the Android phones.

And, you know, we will offload some of that. Um, and I do think that we will probably not have such a, you know, corpus of AI that right now open AI is sort of the center of the AI model universe. And despite their, you know, billions in money and, and partnership with, um, Microsoft, it's still a very centralized amount of AI compute going on there.

I know when I was a child, my father would come home and regularly yell at me about leaving the lights on. So are we not kind of getting the same thing with AI prompts? It's gonna be like, how many prompts did you need and do you realize how much energy you could consume?

Well, I, ideally, ideally that will be automated by ai. So you'll know, uh, the, the, the right amount of prompts and maybe you'll be more conscious of it. But of course, if there's a reminder, yes, that would definitely help.

But I do think that there are ways that are not, and I wanna represent this to the panel as well. I think that there are ways that, um, can be automated, can be optimized that we probably are right under your nose, but you're not realizing it. Like, for example, the one I, the example I used about time of day, um, energy use, what, what can be automated?

What resources are we pulling right now that maybe we don't use very often that we can put a little less energy towards? I think that for, for maybe not the largest companies, but maybe the ones that haven't really brought this to the forefront, they may find that this finops and green ops combination will be able to, uh, force them to look at it from a different lens where maybe in the end, which is the end result of, of saving money. All Right.

So I think, you know, we've been running along on some of our other sessions and I think, uh, we kind of got to the point 'cause I think you're spot on. It's about the metrics. Yeah.

And the, the more you can at at least see what's going on, the, the more your behavior will change. Even if you don't believe in, uh, whatever it is that you're being measured on, uh, if other people do, you'll change your behavior. If you're given the metrics, right?

If somebody kind of tells you something, you'll at least think about it. And that's about as much as we can hope for at the moment. Hey guys, thanks for being on the show.

I want to plug our next segment here is with, uh, Bob Ruman, who is gonna be, well, you know, Bob, he is issues, and he's looking at the fact that, well, do you really have choice when you only have three choices? Or how many choices should you have? And a lot of times, you know, we confine ourselves even in the age of the cloud.

We have, uh, t-shirt sizes, small, medium, and large. And does that just kind of narrow our thinking? And so Bob's gonna kind of open your mind a little bit, I hope.

And after Bob comes the rest of the lineup for the Techron TV gang today, or the Techron TV series, I should say. And, um, by all means, check that out because, well, it's gonna just be more awesome and riveting conversations. Thank you all for spending some time with us.

We'll see you next time. Okay, so this is a story about choice. And it goes like this, it seems that we live in a duopoly or ly sort of world where we have choices between two or three things.

So for example, we have like Amazon or Walmart, or Coke and Pepsi, or Fender and Gibson, uh, Intel, a MD, FedEx, UPS, uh, Mac, windows, or maybe three Choices, AWS Azure, Google Cloud. Uh, it makes sense that things consolidate because you really do need huge market share in order to stay in business. Uh, the side players are just the side players.

It really is the big companies that can command all homeless, monopolistic market shares that can succeed or even survive. So it seems that our choices and things are a little limited. Now, that's not to say that we don't have broad choice.

For example, in the database world, we're seeing a lot more activity because Postgres is taking on a commanding market presence, whereas before it used to be Oracle or MySQL or Microsoft SQL Server, there's, there's a lot more, uh, diversity in a database world. But in technology or in general consumer stuff stuff, it really is duopoly ly. So what does this mean?

Do we really have choice? Well, yeah, we can choose Coke or Pepsi. And what's our criteria for Coke or Pepsi?

Oh, I like the red can as opposed to the gray can. Who knows? But it's not a lot of introspection, which leads me to this point.

Once you start reducing choice that there's only one or two or maybe three things to choose from, you might really start losing the ability to analyze things in order to make a choice. Um, if everything sort of looks alike and they just have different names, what real choice is there? Why do I have to think about it?

I'll just go with whatever color I happen to, like, and that sort of degrades the decision making process, particularly in light of other things such as maybe Tory and Labor and Republican and Democrat. Uh, maybe what's the real big difference? And there's arguments to be made there, but the fact is we don't have a wide variety of choice, and that limits our ability to think about making choices.

Does it matter? Uh, I don't know, but I think it's something worth thinking about.