Peter Guagenti - The AI is an Iron Man Suite for the Mind

DevOps Shorts, DevOps Shorts, the show to listen to when your DevOps hurts. And even when you're going strong, it's short and sweet, so come along. Hello folks and welcome to another episode of DevOps Shorts, or should I say Platform Shorts, the show where I invite wonderful human beings to have a lightning-fast conversation about devs, ops, and other mythical creatures. And today's episode is special because we're not going to be talking straight platforms or devops, because I have a very special guest with me today.

It's Peter Guagenti. So Peter is a president and CMO at Tab9, the AI coding assistant, the all-language AI coding assistant, right? And before that, Peter, you were at Cockroach Labs and at Nginx and at Single Store. So you have a lot of experience in the open source space and specifically in developer and platform tools.

So you do have a deep understanding of our field, and I'm very happy to have you here. Thank you, Anton. And without further ado, we'll go straight on our three questions. And the questions today are also a little bit special because we want to talk about, of course, Gen AI and how it relates to our field.

And the first question will be about that. So Peter, how do you already see AI changing our industry? and what AI tools do you already see applied in specifically the field that interests my audience, which is platform and DevOps domain? Yeah, well, generative AI is changing everything.

And I think it's easy to think that it's all hype because we go through these cycles every three or four years in technology where something is the new thing and it's changing everything. But generative AI and AI broadly, which has been building, for call it seven to 10 years, you know, we've seen algorithmic applications on the rise. I think we hit a tipping point where generative AI has embedded itself in how we work in a way that is here now, right? And my belief on this, having lived through the birth of the internet and started my career building web applications and websites in 95 and, you know, having gone through multiple waves of that network birth of cloud, this is as big as the rise of digital.

Like this is truly as big as that. And with regard to how it's changing how we work, I would argue that actually as technology professionals, we're at the leading edge, right? The key here has been that the LLMs, which allowed this to be possible, they work really damn well when a language is very well known and very well structured. And, you know, something like the English language is a problem because it is a mess as a language.

But programming languages are not. They're organized and they're orderly and they're structured to them. So if you actually look, Tab9 created our first AI coding assistant five years ago. Copilot came out now two years ago.

You have this momentum building, and we're starting to see a pattern of how we actually work if we have one of these tools living alongside us. And the impacts are staggering. I look at all the third-party independent research. I try to avoid all the vendor research because it's always going to be slightly biased.

but you look at like what Carnegie Mellon and McKinsey and IBM research have all shown and they're all seeing roughly the same thing which is a 20 to 25 productivity boost for developers across their entire set of things that they do You know you seeing repetitive tasks and things that are sort of mundane lightweight tasks reduce up to 80 You know, and I think that's not going to escape the DevOps realm, right? I mean, there's so much of what we do that is repetitive. There's so much of what we do that is rote tasks that we're working through.

Or once we have a vision and we have a structure and we know how to do it, you know, we don't want to sit there and just type away, right? We want support to be able to generate these things automatically. And the tools have gotten to the point of where that's absolutely possible. And I think if you take the mix of generative AI, you know, you think about infrastructure as code and all the work we do there and the impact of generative AI can have on that.

And then other forms of deep learning and machine learning that we use to automate certain scans and studies or doing things like vulnerability scanning, et cetera, in our normal workflows, then I think if you fast forward three to five years from today, our jobs are going to look nothing like they do right now, just like they didn't five years ago. Great. So, yeah. So, it changes everything.

I agree. i guess uh the the actual practices are the things that we still need to to discover because right now i suppose everybody does more or less of the same i don't know goes to to the chat gpt and ask for a piece of code or maybe uses a coding assistant and lets that complete whatever they're doing right Yeah. Yeah. I mean, I do think there's already changes that have happened and your audience can start embracing them.

Like, you think about what we do around building configuration for deployment. Like these configuration files are things that are very well known, right? And with the AI coding assistance, you can give plain language instruction. And if it has the context of what you're talking about, like a configuration file, it'll just generate the file, right?

And so, you know, I think the old world of where we had to plan, we had to design, then we had to, you know, implement through sort of those repeatable tools, then we had to go execute and do all that. That's already changed, right? We're more focusing on the plan and the design, and there's an acceleration in the execution. I think what starts getting really interesting for the future for DevOps is what happens when these AI coding systems have access to everything happening?

What happens when they have access to log files? What happens when they have access to alerting, you know, all the observability tools? I think that's where it starts getting really interesting. Like if you go into tab nine today and you give it a log file and you ask it to explain what's going on and what's happening, it'll tell you.

It'll tell you what the errors are. It'll tell you what the actually underlying issue is, right? And I think that's when the way all these generative AI systems is, it requires context, right? So for us, as an AI coding assistant, we're trained around, you know, just our sphere, right?

Just around software development, right? But if you take it that next level and say, okay, I'm going to give these AI coding assistants access to all these other tools, they add that as context, right? It's think about memory, right? It's they now understand more of how you're working, what you're doing.

So if you ask them to explain something, the parameters are so big now on how much information you can feed it and how much it will explain back to you. I think a lot of the time we've spent as platform operators is trying to scan across five, 10 systems to understand what's happening and stitch together that insight and come to some hypothesis of what the issue is. But AI is going to do that so much faster than we ever can, right? We still have to know what to point it to and what that means.

And we have to understand what the behaviors are. But I feel like there so much of our work that art and science and then there so much of our work that just sweat equity I think a lot of the sort of sweat equity goes away and we get to focus instead on okay what are we seeing What happening And then what do we think is the underlying cause And then working through that. With this answer, you skipped a little bit to my last question, which is usually about the future. So you're, but I have something else to ask you about that.

But right now, the most burning, troubling question that everybody's asking, is AI coming to steal my job? Look, it's a fair question because so much of this stuff is disruption of how we work. And my personal opinion on this, and I think it's borne out by what we've been seeing in the tooling, is it's not coming for your job. And it's not coming for your job for a few reasons.

First off, I hate that they call these things copilots or pair programmers, they're not. They're recommendation engines. That's what they really are, right? These are things that will generate material and provide insights based on what you give it to understand and how you work with it.

And so I think about these tools no different than sort of the evolution of, I remember writing code in a text editor. We'd never look back to that. We use IEs now, right? You know, I think it's going to be the same way.

We're going to look back 10 years from now and say, why did I hand type every line of code? Why did I go and do all this? I think, you know, you're seeing this thing really be not a co-pilot, but we talk about it internally as an Ironman suit for the mind, right? That's really what it is.

It's really a way to accelerate the skills and capabilities you already have. Now, to be clear, it's also true if you're bad, right? So, you know, if you're an inexperienced programmer, if you're somebody who doesn't really understand how to construct the application correctly, and you start asking the coding assistant to generate code for you, it might also generate things that aren't going to work very well, right? So I think there will always be a place for people who understand computer science and software engineering and understand how these systems work and how to operate them at scale.

The work though is going to shift. It already is, right? And it already has for so many years. I think the other part of this, if it makes your audience feel better, is a sociological one, right?

We, according to the, basically the federal budget office in the United States who tracks us, the Department of Labor, they believe that software engineering jobs in general are going to grow nearly 30% by 2030. We're not making 30% more people. So right now, if you actually go, even with the layoffs that we've had from overhiring some of these tech companies, the reality is most companies still have at least 10% of their total engineering tech companies open roles. So I think what we're seeing here is actually these tools are coming in at just the right time to help us all deal with the tech debt that we have, to help us all deal with the increasing pressure on us to build even more applications and even more features.

because I've lived on both sides of it as an app developer and as the business owner pushing on these things. And I'll tell you, if you gave me 20% more productivity from my team right now, I wouldn't fire anybody. I put 20% more work through them, right? That's what I would do.

Because I think that's the world we live in. And I don't think we talk enough about as we keep building more of these applications, by the way, the maintenance and tech tech goes up. So there's even more there. So even as we're building, we're forgetting that the things that we built already don't go away.

So what are we doing about it? So I'm bullish on this. You know, I think the jobs are going to change. We don't have typing pools once we got computers and word processors, right?

We don't have, you know, secretaries walking memos around. We have email. Like there will be changes in your job and you have to be prepared for that But I feel like as technology professionals God man like my job today versus in the mid 90s when i started doing this stuff it i been through like four careers at this point it doesn even feel like i've even done the same work yeah definitely okay and finally you know well this whole episode is about the future but still uh how do you see uh gen ai and ai tools uh influencing our creativity because i always like to say engineering is a creative professional so we were talking about okay this will help me with you know the mundane the boilerplate but will will that impact also the way i'm creative with my code i think it will i think it will and i love that you say that i i actually started my career on the on the creative side of house and user experience and that, and also wrote code.

And so I've always sort of lived in both worlds simultaneously. And I don't think software engineers give themselves enough credit as creative people. You are a creator, you are a maker, like that's what you actually are. You're, you know, some, some are more craftsmen, right.

Than artisan, but, you know, I still think it is a creative endeavor. And when you're, when you're creating an application, you can just write in basic functions and adding capabilities. The first thing you do is, is, is be thoughtful and design what it is you're trying to do. I feel like so much of soccer engineering is 20% vision and 80% execution, right?

And so we only get to spend a small percentage of our time on the stuff that's actually truly enjoyable, which is designing and crafting this system that we want to work. And my belief and what we're seeing in the patterns of usage already is that generative AI is going to shift that calculus, right? Instead of 20-80, it's hoping, I hope in 10 years, it's going to be 80-20, right? Where you spend more of your time designing the system and less time actually going and having to think about the minutia of execution.

And we're seeing it already. Like we've got experiments in-house where we're doing full generation of applications in React, like mostly UI stuff, but all based on prompt, right? Just all based on prompt and expectation and it's doing it. Like we're writing whole applications from it.

So think about a world where as a software engineer, you know how your system works. You know how to craft this. You know the components you want to stitch together to create this application. And instead of having to sit there and handwrite each, network each, you know, make everything integrate the way you need to, you can just go and create the design, tell the system, and then test and optimize, right?

I think that's going to be a somewhat more enjoyable work than slogging through writing code. I mean, I'll give you an example at the same time. Like when I started my career, we didn't have content management systems, right? One of the very first web applications, websites I built was the Super Bowl site for Super Bowl 30.

It's been a long time ago, showing my age but you know uh sports fans are statistic nuts like they want peter i hate to stop you but we're done this is 15 minutes for a reason because you know great delivery comes in small batches that's that's our slogan i thank you very much for doing this and And totally a lot of very useful information here, exciting stuff you're working on. Thanks for coming. Appreciate it, Anton. Have a great day.

Short and sweet. Thank you for listening and watch out for new episodes of DevOps Shorts.