Project Wisdom – Richard Henshall, Red Hat

This is Textron TV. Hey guys. Thanks for the throw.

We're here with Richard henschel who's there? A product management for ansible at red hat and we're going to be talking about natural language processing Ai and the future of it. Hey Richard, welcome the show.

Thanks very much, Mark. You guys kind of launched this a little while ago and I guess everybody now wants to know how far along are we in the age where can GPT is all the conversation these days people want to know well just when we think is gonna happen is all this stuff gets applied to the way we manage it not just everything else in our lives. Yeah.

Absolutely. I was very timely wasn't it the Thailand the last year there was a massive surge in the the interest around this topic in your chat GPT certainly helped. So like USA, we announced we announced for wisdom.

You know last year ansible first in October as he's very much. How do we bring the power of AI into the creation of ansible automation content for users and I think you know a lot of people had a you know, why? Why would I want this?

Why would I have this? I think nothing else chap is shown exactly why people want this, you know what the part power and the possibility is and the way I often. Think of these things is you know, we've spent many years thinking about how AI could replace what the worker does what the user does but actually now the power of the capabilities is really about how does AI augment what the user can actually achieve and what they can do.

And this is what we're trying to bring to ansible because you know, there's a natural fit, you know ansible is a language there for purpose for a task. It's a simplified language is very natural in its way that you interpret it. You know, you'd say I would like answer to do X Y and Z right to make my life work.

That's how you ask the question when you say hey automate a VM to build this in this location that this side, you know, you're doing the same the same type of activity. It's a natural fit for how those two work. And we also have a big community of users that rely on ansible to help and you know, do those simple tasks.

We have the large-scale corporate users that use the automation platform as a method for for large-scale automation practice, you know across their orgs and you have a single users that just want to help, you know build something every once in a while. All of those users can gain the benefit of how you know, natural language processing in an AI model can can make their lives easier. Am I going to type in my request how exactly will I interface with this or will there be more of a voice experience where I'll just come in and say, you know Spin Me Up 14 virtual machines and this particular platform and everything will be taking care.

I think the Foy side day would be it wouldn't seem nice in a film but I think it's probably not practical for the way most people work. I think you know the voice assistance have probably taught us that now very much more type and how to use and you know that the initial approach is very much getting into the workflow of the user so through this or developer IDE type experience, you know in typically writing ansible, you know, you type out we always try and say, you know describe what the task is that you're trying to achieve or by describing the task. You're actually effectively asking the question to the model to say this is the automation I would like to do and I think it's probably working out the idea around what project wisdom is and what you know answer all wisdom will be is is more than just the engine there's more than just the language process and the algorithm.

It's the experience around that it's about how you bring the totality of what I user wants to try and bring into the mix so where they get to do it for so yeah as an API that you can talk to what about the testing what about the validation? What about bringing the fine tuning of the model so we can actually attribute where the content comes from, you know, is it the content is ordained by the system administrator in the company? Is it from the community user?

Is it from a partner? Is it from a technology provider? You know did Cisco or F5 or IBM or whoever write that automation clearly trusted in the way you want in some ways that some existing models.

Don't quite allow. To the writing of the automation historically at least it's always felt like one of the challenges organizations had is the person creating the content was generally a specialist. They're often too busy to create the content and they had this domain level of knowledge.

Are we getting the point now where it's also easier to create the content, but the whole thing becomes more accessible to a whole class of it professionals that otherwise may not be the quote unquote high priest of 18. Well exactly that I mean that I think you know, we've got past the point where the sex that's the center of excellence model is the right model to take that's always going to be a bottleneck and that's always been one of the powers of answers boys or how to get that tooling into somebody's hands much more easily don't have to know a programming language. Of course, if you know a programming language, you can be much more powerful with the tool than you you can't without it, but you still have to have that level of access.

And so we've seen that but you still have to know what you want to do. Right? And I think this is being one of those areas where I often people know what they want to do, but do they know how to make it happen in the tool in the most efficient way and that's why I said that comment earlier about it being about augmenting the user rather than replacing the user.

It's really about saying I really know I want to create a virtual machine. I want to create AWS. I want to create it here.

I want to do it there. But I don't know all the syntax yet for how to do that. So, how do I get them there faster and maybe at some point they move beyond the model or maybe it just helps them become two or three times as efficient as they would be so that I really think that the, you know emphasis on the augmentation of the user and then give it helping it to make that everything more accessible to a wider scope of users is a really big focal point when I get to Do you think we're approaching a crisis of complexity?

Because when I look at things I see we have all these Legacy monolithic applications that are being lifted and shifted into the cloud and they still exist and then we have all this Cloud native stuff coming along which is not nearly as easy to manage. And on top of that there's multiple flavors of that. So we have you know, containers and serverless Frameworks and was awesome coming down the pike.

So are we kind of look at something here where you know, this could wind up being too much of a good thing. At a certainly possible, right? I mean obviously when you move into one of those new areas you you know people tend to fully invest and they go in and they leave behind those old things but some of us I use is my my background before I joined Red Hat, you know as a customer and I you know had that architecture role we were looking at all of that complexity and it's always it's always gonna happen.

I think that's always been the case. Yes, there's there's always seems like there's more complexity going forward nevertheless. I think that's probably the same way for the past 40 or 50 years right in how these things are moving forward.

What if we could help tame that complexity? What if we could say well yeah just definitely things that need to move ahead at lightning pace and off they go how do we bring those the older things? How do you mean the things the middle?

How do we help people get into that place? Faster right in a more efficient way, how do we optimize that experience for them? And I think that's where this opens up a new opportunity for us to all that also think.

You know, how do we supplement those people? How do we help with that complexity because it isn't going to go anywhere. How do we make people successful with what they've got and that's really why I think that's all has a powerful play and there's other or the tools in the same space with a lot of us thinking about the same stuff.

It's all about that uplifting that same complexity that we've got with the new the new capabilities. Well, how do we take those capabilities to apply them back? To the older capabilities that we don't leave them behind because most of those older systems.

They're the ones that are driving, you know, the actual core business intelligence that most organizations run on you can't just replace all of those. You've got to marry that with you and Pull It Forward and it has to be looked at someone has to look at it holistically right as a group right while we let those others go off and strife and you while we all help those others maintain the old so this there's that place to play in the middle. There's about bringing those two areas together as well.

Things never seem to go away. I'm oh one should people expect from AI as far as it is applied the it operations as we go forward. There was a lot of initial hype in the category and then there was skepticism and now people are kind of seeing all these other models.

So from your perspective, you know, looking into your crystal ball for the coming year. What do you think we're gonna see and experience? Well, I said a big chunk of it is again, is that how do we add right to what people do how do we make how do we make their lives easier?

But I think it opens us up. To a bunch of other other areas now wait, I've taught, you know here very much. This is about how do I get bigger broader a bunch of people that maybe don't know what they need to know and bring them together and help get them on board.

But then so I think about well. Automation is there and you know, the operations tasks are there to deal with the routine the repair the repeating repetitive tasks that we always have to try and Achieve. How do we take them out of the way?

As you make it more accessible and easier for those things to be applied then we also start to broaden and widen. What we see as opportunities to have that same pattern applied for and so I do think that the you know in the the move up into the operations is you've got two two fragments to AI you've got the creation of what you want to achieve what you've also got Observer. This is the analytics I think actually what we're doing with the creating automation that becomes a really powerful partner to the other the other existing AI that exists around observe Observer that ability monitoring analytics, you know, detecting something that we think is wrong.

Well. Okay now I can pass that instruction. To A system that could potentially automate what the answer is that it wants to be to either respond remediate discover what that can potentially be.

And that's something we're starting to look at as we start to combine some of the other areas around sport that we're doing around event driven automation, you know combining what wisdom might be able to offer. But you know think about the the other side as well the ongoing maintenance the ongoing planning for how we do that work, you know, once you it's great. It's easy to write a thousand automations, but you're gonna maintain them perpetually for you that those things also don't go away.

Can the AI model be brought into that space has it help us maintain the Automation and the code that we've created. How does it make that part of our Lives easier so we can still keep focusing on the things that add value and taking away that sort of technical debt and churn that consumes our time as years go by How do we govern there as one wagon said it's one thing to be wrong. It's another thing to be wrong at scale.

And with AI we can be really wrong at scale. So how do we kind of get in there that process and have confidence in it or be able to do something that may be in intervene when we need to yeah. Well my background will give away by the statement but there's still separation of Duties right in some of these things in my in my view of this, you know, it's not all encompassed in one system.

Right? We have to be about bringing those different pieces together. So I want something to detect and understand our need something with that knowledge.

I also have to have something that's there for my security from a secure perspective governance governance perspective something that can be audited and again this, you know this mix of those two having something that takes responsibility for that side of the organization because that's that's what a lot of the it operations teams. Historically always did it was about making sure we had continuity of As we had the quality and the expectation the release the change management, but automating into that space, you know has to also be done if you want to automate and add all that value. To the so if the change agents of business where you know, we're more Revenue we made were busy outcomes can be achieve.

So marrying those two together right by having good systems that integrate. Well that can talk to each other that can show that so you can build the trust prove the trust, you know, add decisioning capabilities in add authorization capabilities in you know, so that that will be the key to really making these things come together in the right way. Years ago we spent a lot of time worrying about the number of it administrators per virtual machines or whatever else that we were measuring out there.

But it seems like we can now manage it at a level of scale that's not really tied to how many people we have. So, how do I know whether I have the right level of it investment per what measure these days or is that just the whole Antiquated conversation to begin with? I think it's going to differ based operational models, but you know we're seeing I think the more important thing is the skills Gap.

You know the number of all the number of Administrators you've got I think is is less important than if you've got the right type of administrator with the right skills to achieve what you want. You know, I think I was doing something else by this recently and you know devops has been very successful for many people. It's also not being successful many people.

And I said so it's both succeeded and failed But ultimately it's where we failed. It also showed that other you know, there are people in that space, you know, the reasons why devops came about that friction between the development space and the operation space the infrastructure teams. Came about is shown that there are people in that infrastructure team that can go to that next step and then you get into the world of the platform engineer now and the SRE function.

And so I think having the right skills in those teams and the right mindset for what you're trying to build and why you're trying to build something becomes more important than the number you have. Of course, you have to have enough right for them have work you want to try and do by saying, you know, I've got a thousand to one or ten to one could be a bad ratio and a thousand wanted to be a good ratio is the right skills exist in those individuals to automate correctly scale, you know by the right governance of our process right security posture all those things. And so I think the skills become the important part and I think we're seeing a lot of challenges about the skills gaps, we've got so if again segmented back to my earlier point of how do we bring all those users up while succeeding in letting the other groups achieve what they're trying to achieve well, All right.

So we're heading now in the 2023. What's your best advice to folks? What's your crystal ball telling you for the year?

It's there's gonna be a lot of change this year. I think it's gonna be there's a lot of external factors. They're gonna make people have to think about what they want to do.

It might be challenging depending where we are. I think there's opportunity with that as well. I think there's gonna bring a focus and attention, you know efficiency optimization is what I'm starting to see happen for people in in my area and automation space, you know, what is the power there?

Yeah getting getting that value from what you're able to do right bringing those skills to the fall scaling them out making it possible for everybody to gain as much Advantage from that skill set, you know skills are there to be knowledge is there to be shared right knowledge becomes something that we share with others right becoming guidance to those others. How do I scale that guidance isn't by working Harder by working smarter. And so All right, folks you heard in here as always changes the only constant.

Hey Richard. Thanks for being on the show. Thanks so much Mike.

Thanks for having me. All right and back to you guys in the studio.