Kyle Campbell, CTO.ai | KubeCon + CloudNativeCon North America 2023

This is Textron tv. Hey, everyone, we're back. We're here in Chicago at CubeCon.

We're wrapping up our day two coverage, our last guest for day two. We'll be back tomorrow though. But let me introduce you to Kyle Campbell, right?

That's right. ai and a little bit about what we're they're doing here at Q Con. But before we do that, Kyle, first of all, welcome.

Thank you. Second of all, let's hear a little bit about Kyle. Yeah, thanks.

Tell us your kind of journey. Yeah. Well, first of all, great to meet you in person.

I've talked in the past. Yep. Yeah, so, so my journey, I, I've told this story before, a little unconventional.

I grew up in Nova Scotia in Canada, a small town. Yep. Been on the internet since the age of eight and just had no interest in the formal past.

com boom. Uh, I may not look at it, but I got some of the scar tissue. Yep.

And I, but I was self-taught software engineer, so open source was like the key to my success and Sure. Good developer tools. And, uh, I came up through the cloud and open source era and then started founding developer platforms in, uh, 2014.

The, the first company I built, uh, was a developer platform, the real estate space. Okay. Zillow acquired it in about eight months, which was interesting.

Very. Um, and then I, uh, I bootstrapped a, a DevOps agency quite successfully and, and started to find that there was, uh, a lot of opportunity for next generation developer platforms, which led me to CT ai. Excellent.

Man. I, what a great story too. Yeah.

Still up in Nova Scotia. I'm not, I moved to the other side of the country. I live in British Columbia now.

Good for you. Columbia Outdoors. Spent a lot of time with my, my son camping, fishing, and, and trying to get outdoors and just enjoy The be.

Yeah. No, it's beautiful. Beautiful.

I mean, not that Nova Scotia's not beautiful. Yeah. It's brutal in the winter.

Yes. But it's, but it's beautiful country. Yeah.

Part of the country. Um, let's talk CTO AI now. Sure.

So, look, every I I've founded multiple companies myself. Every founder I've ever interviewed or spoken with in 30 years, they don't just say, wake up and say, oh, I feel like find founding a company today. Yeah.

There's, there's kinda like Richard Dreyfuss and close encounters of the third kind. Right. Right.

There's something driving you like, I gotta do this, this needs to get done. What was, what was driving you that needed to get done here with CTO ai? Yeah.

I mean, as I described my past, a lot of my journey was self-taught and, uh, stand on the shoulders of giants. And really important thing for me was developer experience and ease of use and tooling early on in my career, because that enabled me to really, um, drive my competencies as a developer and keep up with these people that had, uh, you know, computer science degrees and master's and all these things. Right.

So, uh, when I thought about that, and I also thought about where the future's going and my, uh, passion for platform engineering, uh, it became pretty obvious to me that, um, what was missing in the market was, um, a set of tools that were designed to embrace the future of developer experience and how AI can impact that from code to cloud. So, um, in 2016 or so, I started experimenting with the idea, and, and through, uh, 2020, I decided to formally launch C two I as a, as a developer control plane, is what we call it. The way to think about it is it's a modern CICD platform where teams can compose their workflow across build, deployment, test, and preview.

And as they work with these tools, they can interact with them in the CLI, slack, GitHub, talk to these workflows through sort of a prompt based model. And we collect telemetry data that allows them to then measure that against Dora metrics, which is, um, you know, standard. Yeah.

It's a standard. And, and, and, and I think frankly, an empirical way to measure the developer experience, we don't try to measure developer productivity. We think that's the wrong question to ask.

The question is, how do I measure the experience the developer has in the same way that I would measure the experience my users have with my product? And if what we're doing is building platforms to enable developer experience, we need a natively integrated way to measure that. And we also think that that measurement is a way to pave a future, um, between how the developers experience those tools and how AI can support, um, the developer experience and enhance developers with frankly, superpowers over time.

So, you know, I, I was recently at a couple of user conferences, big DevOps companies, and it was interesting, three conferences all use the same metric, which is that developers only spend about 30% of their time Yeah. Developing State of DevOps report 2018, I think. Yeah.

Yeah. And, and, but it's been, uh, it was in there and it's been kind of verified in several others. Yeah.

What do you think about that? Is good thing, bad thing, something to be proud of, something we should be working on? What's the deal?

Yeah. Well, I, I, I think a pretty practical point of view in that, um, as a software engineer, my goal is to release features that customers love and drive business value. And, um, you know, if I can do that at a, at a higher velocity while maintaining quality, I mean, I, I think that that is a way that I as an individual can contribute to the success of the business and the team.

Um, but I think what often happens is we get really obsessed about the individual developer. And I think any developer who looks at the idea that they only spent 30% of their time adding value, um, isn't gonna like that very much. And I think we need to reframe the conversation away from developer productivity towards, like, developer enablement and look at the tooling and the process that, uh, the developer experience promotes and how developers go from code to cloud, um, and ask ourselves as a, as a business leader, as ACTO, uh, who we work with a lot, like is the tooling and the workflow that we're supporting our team with and that developer experience, what they're experiencing is that driving results.

And I think door metrics is, uh, an empirical way to measure that, obviously through anecdotes and surveys. But like, what we've really innovated on is creating a vertical integration with the tool chain that allows you to measure that in real time and then iterate on the tools quickly. And some of the teams can really own that themselves.

And what the businesses should be doing is, is zooming out and asking themselves, yeah, how do we maximize the amount of time developers spend without, you know, individualizing that consideration. There's an interloop and there's an odor loop. And what Dora's really good for is measuring the odor loop of how we work in aggregate as of teams.

Um, we need next generation platforms that make this data and this context more accessible to business leaders. I, I, I don't disagree with anything you said there. So we we're there with that.

Um, what about, have you guys released any news or anything? What do you got going on around the show? Yeah, so for the last while, we've been focused on really like building a rock solid, um, you know, workflow and tooling, uh, to basically like set the premise of that next generation composable CICD vision.

We've fed that data into Dora Metrics. You can now report on Dora metrics in real time. Save those reports, share them with whoever one-on-ones retros kind of up to you.

And we're starting to experiment with some data science, but something that's really exciting that we just kind of, um, put into production last week, we haven't really officially announced. So I guess this is a first is, uh, we've released our first kind of, You heard it here first. Yeah, it's Ahead here.

First. Our first ai, um, feature set, which is, uh, AI code review. And, and what we're doing is we're letting teams start with these sort of base models for how they can leverage, um, AI in the code review process to get developers feedback earlier.

Uh, but then we want them to be able to do is enhance that model to really, um, customize it to their preferences on their own kind of hardware, um, to customize those models to take into consideration their own specialized requirements and obviously protect, um, the proprietary information that's so core to their, their beliefs. So we launched that into production, uh, announcing it here first, really, and we're looking for early adopters who can kind of test it out and help us stress test it and figure out how do we, um, further enhance the context of that model with, um, some of the information that is derived in these Dora metrics as well. So here we're talking to this camera Yeah.

Developers out there. You're looking for them to kick this around a little bit. Kick the tires.

That's right. Yeah. How did they do it?

You're a developer, uh, and you're interested in how AI can support your CID workflow, specifically code review, come over to C two I, uh, sign up, send a message. I'd love to talk to you personally, software development yourself, and I'd love to hear what you think, what you think are the limitations of hesitations, and more importantly, like how do you think we can enhance this with other use cases? Because I think there's a whole exciting, uh, set of use cases for AI and software delivery that is gonna help software developers.

Um, Absolutely. Yeah. We embrace it.

You know, I, I had this conversation in an earlier interview today, look for, for people who are in tech, you show them, you know, the, an AI interface and some of the kind of parlor tricks it could do, and they think it's magical. Yes. You Know, Arthur C.

Clark, right? Yeah. Now we know it's not magic, it's just putting one word in front of the other or one, you know, but there are some uses that AI really shine certain, so it's not meant to replace Google search, nor is it meant to replace software developers.

No. But it can empower software developers to be 10 x more effective. And one of the ways it can, it could teach you to code, right?

You could put code in there and say, explain it to me. Yep. It could review code.

Yes. Right? So either it's writing code or it's reviewing code.

You shouldn't have it doing both probably, but it's true. Right. But if it's using it for review, what a great use of the technology.

Yeah. We've seen that, like the number one thing that slows down time to market and feature sets, um, for developers is waiting for code review. And while I'm not sure that I, it'll be the same thing as, you know, you and I pair programming and getting into the nitty gritty details.

If we can make sure that every developer can get a code review instantly just by adding a label, and then that we can tailor the quality of that code review with more context that is supported by how the team thinks about quality. Yep. Um, I think this is an important example of how we can accelerate, uh, the needs of developers and the developer experience with AI that still leaves all the control in the hands of the developer to decide, is this rational?

Is this a hallucination? Is this the appropriate time? But the, uh, feedback loop that we can get from this is, I think, a really great beachhead and, and why we chose to start there.

Absolutely. And, and we'll see, I, you know, I don't think we scratched the surface yet on how we're going to use ai, but this is certainly a, a real definitive first step that I, I, I see us going with, right? I think so.

And, and I think the call to action for other software vendors out there is to think about like re instrumenting what they're offering to, to prepare for this 'cause because I think there's a, a necessary hesitance around ai. I think, you know, rightly so. It's, uh, but if we can, you know, think about how, uh, we can take control of it, how softwares can take control of it, software engineers are perfectly suited to be some of the like, first stewards of these problems.

Just like we see AI and code generation. Um, you know, I think the next natural step is to support it within code review, and then we'll start looking at how, um, smarter deployments, smarter rollbacks, and obviously operations can be enhanced with ai. And, and this place is a perfect example of people trying to solve that problem.

Absolutely. Absolutely. And, and here's the other thing.

I've been in technology a long time. You don't put these genie back in their bottles, right? Once they're out, they're out.

And, and so like it or not, they're here. And I think you could, you could, if you can't beat 'em, join them. Yeah.

You can't, you can't for stop Moore's Law. I mean, This is, you're not the pathway That the world's Not, you're just not gonna stop it. That's the way it is.

It's great stuff, man. I love, I love the attitude. Yeah.

Thank you. ai. Again, if you're looking to help Kyle out, kick the tires on this AI code review, uh, product they got coming on, it's fantastic.

Yes. So let's check it out. All right.

Thank you so much. All right. Hey Kyle, enjoy the rest of your CubeCon.

Yeah. CTO, do AI wrapping up day two in Chicago at Cube Con. We're gonna be here tomorrow morning.

I think we're on at 10. We're only here till two tomorrow 'cause it wraps early tomorrow. But, uh, we'll be back with a full load.

We hope you've enjoyed today. Um, on Demand is available at Techstrong do tv. And, uh, so if you missed anything, do check it out.

This is Alan Shimmel. We're out.