Spiros Xanthos, Resolve AI CEO on AI Agents for Production Software

Grace Gong (0:00)

Hi Spiros, welcome to Venture with Grace.

Spiros Xanthos (0:02)

Thank you.

Grace Gong (0:04)

So I'm so excited for you to be here because you just spoke at our conference, and obviously people loved our piano.

Spiros Xanthos (0:10)

It was a great event. Thanks for inviting me.

Grace Gong (0:13)

It was, I feel like, okay, so I want to start with your background. Where are you from?

Spiros Xanthos (0:17)

I grew up in Greece.

Grace Gong (0:18)

Greece, okay.

Spiros Xanthos (0:19)

The most beautiful country in the world.

Grace Gong (0:21)

Oh, I know. And then you came to-

Spiros Xanthos (0:26)

Then I moved to Urbana, Illinois.

Grace Gong (0:28)

Yeah, I know. I went to school, too.

Spiros Xanthos (0:30)

Yes, we're in the middle of nowhere, as you know very well.

Grace Gong (0:32)

Yeah, exactly. So, well, I feel like that really contribute to your career, because I feel like even though Urbana-Champaign was like the number 3 party-

Spiros Xanthos (0:44)

Are you saying that I was trying to escape?

Grace Gong (0:46)

This is not a rage bait. This is not a rage baiting, but Urbana-Champaign is like number 3 party school in the country.

Spiros Xanthos (0:53)

Not if you're a PhD student.

Grace Gong (0:54)

Oh, okay. Yeah, I feel like that.

Spiros Xanthos (0:56)

You did your undergrad there, right?

Grace Gong (0:57)

I did my undergrad there. So, yeah.

Spiros Xanthos (0:58)

Maybe it was for you.

Grace Gong (1:00)

Well, no comment. Okay, so I want to start with like, how did you come up with the idea of Resolve?

Spiros Xanthos (1:09)

Yeah, so I previously built actually multiple tools that developers use to monitor production systems. You know, I've co-created OpenTelemetry, which is now the de facto standard in how you collect telemetry from systems, which then every monitoring tool uses, Pentalog Analytics Company out of my PhD from the University of Illinois.

And it became apparent to me when I was at Splunk and managing a very large engineering team, that the biggest pain at scale is actually how do you run and debug these production systems. And despite, let's say, unlimited access to our own tools, because we were an observability vendor, it was still very painful. We had a period of time where our entire SRE team resigned because of burnout, because of how hard it was to keep the systems reliable. And then, having this deep understanding and empathy for the challenge, and then maybe seeing early on the possibility that, let's say, the models will keep evolving in a way that maybe the future is not going to be one where we collect all this data and we dump it back to humans, but instead we have AI that does the work and solves the problem.

Grace Gong (2:15)

I want to start with, use one sentence to describe Resolve for the audience.

Spiros Xanthos (2:19)

So Resolve is an agentic SRE that can be on call on your behalf if you're a developer, receiver will alert every page all the way to resolution. If it's a problem that it cannot solve by itself, it will page you, but with a lot of context, an actual root cause and potential remediation.

And beyond that, it's also like a fully, you know, agentic platform for SRE automation, right? It can participate in incidents, you can give it any task you want, but it's production-related, that it will automate. Not different from how you give your coding agent task and code, you can give Resolve any task for production and it will perform it.

Grace Gong (2:54)

For sure. Okay, so I have so many questions there. One is like, let's start with like the big picture, right? Nowadays, like I feel like in the AI realm, like everyone's like coding with agents, and like with, whether it's like Codex, Cursor or whatever, but like, what's your thought on like the broader landscape of like the engineering jobs? Are people gonna be like, keep doing the coding themselves, like since like right now all the AI is doing coding, and then I feel like Resolve is like serving the agents, or how do you picture like Resolve in the entire like engineering stack, and how is like the engineering stack kind of like shift?

Spiros Xanthos (3:27)

Yeah. So, I definitely see a future. We're already living it, right? Like at Resolve, where the vast, vast majority of code is being created by agents. And then the role of the humans on the development side is to essentially set the goals, oversee the quality, set in some sense the longer term, do the longer thinking and strategy about what needs to be built, what's valuable, what the users care about. But that's only half of the picture, right?