Tristan Ratchford has joined Hg Catalyst as CTO. He comes from Datadog, the monitoring and security platform used by engineering teams to keep their software running.

Tristan spent five years there leading the team that built Bits AI SRE, Datadog's flagship AI agent that investigates and diagnoses issues autonomously.

Catalyst is Hg's AI product incubator, launched publicly in November 2025. It embeds expert teams of engineers, product leaders, and designers in Hg’s portfolio to accelerate agentic AI product builds. Tristan's appointment is the latest step in building the leadership and infrastructure to do that at scale.

He sat down with Lloyd Hilton, Head of Hg Catalyst, in his second week. They talked about what drew Tristan to Catalyst, how Hg has been building towards this for longer than it looks, why vertical software is a particularly interesting arena to build products, and how the pace of AI is shaping the operating model.

Highlights below. Or watch the full conversation on the Catalyst website.

What drew you to Catalyst, Tristan?

Tristan: The big tipping point for me was when I started to meet more of the team. I was absolutely blown away by the calibre of the talent here. I was talking to one of our engineers, going very deep into AI, we ended up on topics like AGI, and then quantum mechanics came up. And he said, "oh, I did my PhD in that." These are world-class people.

Datadog is a very special place for me. I worked there for the past five years, and building Bits AI SRE has been the single biggest accomplishment of my career. I got to work with a top talent team and go really deep in a product area. But the opportunity to do that 50 times over at Hg was super exciting. I had to join.

Lloyd, how did Catalyst come about?

Lloyd: Hg has been investing in data and AI for about a decade now. Chris Kindt had the idea that we could build a data team at Hg, and over time that grew. Dr Amr Ellabban, our Head of AI, came in about eight years ago. Our initial focus was improving the data infrastructure for our businesses and building better reporting products.

When ChatGPT launched, we started to explore what we could do with it. Initially we looked at our businesses in industries more at the leading edge of AI, so legal tech for example. In 2023 we did a couple of builds, with Litera and Blinqx in Europe, building out early AI agents and early RAG (retrieval augmented generation) pipelines in legal tech.

The thing that really catalysed it was when Chris, Amr and I spent time out in Silicon Valley, end of 23, early 24. We met a number of the leading VCs who were starting to invest in this trend, and some of the early AI natives that were emerging in spaces like customer support and coding. At that point we had a watershed realisation moment that this was the platform shift. From then on, we went back to the portfolio and started thinking about where the opportunities were to deploy this. Catalyst is the latest evolution of that.

Why is vertical software an interesting place to build AI products?

Lloyd: You can dive in and immerse yourself in property management, then after six months step out and immerse yourself in accountancy. These domains are in regulated industries and don't sound that exciting on the surface. But once you actually get in, there's fractal complexity that sits beneath what you'd see from the outside. Really deeply understanding those verticals is one of the things I continue to find interesting and rewarding.

We start with the workflow, with the customer jobs to be done. We don't start with what our software does today. Consistently we find AI widens the aperture of the value you can deliver to customers. It's about reimagining these verticals. The core system of record SaaS business is just one part of what is now becoming a much broader system of action that we're building for a lot of our companies.

What have you learned about how to make these product builds work?

Lloyd: The first thing that's become more prominent in the way we do things is getting evals in place. I remember our first build back in 2023, we didn't really know what an eval was. Evals are now front and centre of what we do at Catalyst.

The second is discovery. If I look at any of our builds where something has gone wrong, often it's not the technical delivery, it's that we didn't ask the right questions of the customer, or we weren't embedded closely enough in the domain. So now from the start we bring customers into the conversation and bring deep domain expertise into the discovery and build process. We're also deliberate now about integrating with GTM and pricing from day one.

Tristan, what does the speed of AI development mean for how you're running engineering at Catalyst?

Tristan: The startup ethos isn't a recruiting tactic. Given how fast AI is moving, we have to have that quick iteration cycle, that quick delivery. The last couple of years in AI have felt like the invention of the internet, completely compressed down to a year or two. Keeping up with that means we have to have that pace.

The way I plan to run engineering is with that startup ethos, moving fast, combined with research rigour, to make sure we're delivering things that actually work.