AI and LLMs in Engineering: Scaling from Startup to 1,000 Employees with Jon Mort, CTO at The Adaptavist Group

Jon Mort on Growing Adaptivist to a Thousand People, AI as a Coding Partner, and the Power of Connecting Teams Who Did Not Know They Needed Each Other

Jon Mort, CTO, The Adaptavist Group | Interviewed by Gareth Watts of Atolio

Jon Mort is the CTO at The Adaptavist Group, which helps companies get more out of their Atlassian, Monday.com, and GitLab tooling. Gareth Watts, Atolio's own CTO, talks with Jon about his nerdy path through computing, what it has been like helping Adaptavist grow from 12 people to over a thousand, how he thinks about AI's real role in engineering today, and why the most exciting thing about tools like Atolio isn't finding documents but finding the people who have the knowledge you need.

The Nerdy Path In: From Breaking Computers to Building a Career

Gareth Watts (GW):  How did you get started in engineering? Was it a childhood passion?

Jon Mort (JM):  I was a very mathy kid. I just loved solving problems. My parents got an early Amstrad computer that I broke several times programming on it. I absolutely loved it. Then I went to university to study artificial intelligence because I thought that was the future. It turns out I was wrong in the short term and right in the longer term. I went straight into software engineering roles out of university and spent the next couple of decades there. It's that love of maths and problem solving from early on that got me into this world.

The thing about getting kids into computing today is genuinely different. When I was young, the internet didn't exist. You typed in programs from books, you tried them, you edited them. You were very focused because that was literally all you could do. Now kids are bombarded with options: Minecraft, Scratch, everything else. My eldest is big into mountain biking, and I try to encourage him to think about how the mechanical problems he solves there translate into broader problem-solving skills. The key message is always curiosity and fearlessness. Not being afraid to break things is how you actually learn.

Learning by Going Deep: Linux from Scratch, LLMs, and the Value of Building Upward

GW:  Do you think that approach of building from first principles still transfers to things like machine learning today?

JM:  Absolutely. I've gone through some of the FastAI material that Jeremy Howard puts together and the resources that Andrej Karpathy puts out, and I think you get a completely different level of understanding from building something up from very basic building blocks. Seeing the gap between a super-simple thing you can build yourself and the superpower of something like GPT-4 gives you a real appreciation for where the magic and the effort actually lives. It also demystifies it. Once you understand what's happening under the hood, you can reason about what these tools are good for and what you should avoid.

AI in Engineering: Superpowers for Developers, Not Replacements

GW:  How do you see AI changing the engineering function?

JM:  The person who is going to take your job is someone who can use the available tools better, as has always been the case with any new technology. When Google came around, the people who figured out how to use search engines to be more efficient at their jobs did better as a result. LLMs are the same category of shift.

For engineering specifically, what I see is that you're doing a lot more code review. You've now got a coding partner sitting there the whole time generating suggestions. Some of it is brilliant. Some of it is wrong. Some of it is nearly right but has a subtle mistake that you need critical thinking to catch. It's a partnership, a kind of two-way coaching. And I pushed back hard when our finance team suggested it might replace junior engineers. It gives them superpowers. That is a completely different thing.

The critical thinking element is irreducible. Someone still needs to ask: is this the right thing to be building at all? No tool changes that. My advice to teams is to build for the capabilities that are coming, not just what exists today. If you're thinking about what you can do right now or three months ago, you'll always be lagging. Think about the product you can build in six months given the trajectory of improvement. And make sure you're building in a way that lets you swap in better models as they arrive.

Why Adaptavist and Atolio Are Working Together

GW:  You connected with us fairly early. What did you see that made you interested?

JM:  The first contact I had was through Mark, your co-founder, who had heard me talk about information management on a podcast. He reached out and said: we have this thing, what do you think?

The problem he described is one we see with a lot of customers. You have information everywhere. Knowing where to go to find what you need requires a tremendous amount of institutional knowledge. We had actually tried to build a unified search into Confluence for a customer once: an engineering team searching across their codebase, wiki, issue trackers, and requirements system. They didn't really use it. Not because the idea was wrong, but because it wasn't complete. If it doesn't cover all the sources people need, they don't trust it. That completeness problem is exactly what makes the challenge hard.

What excited me about Atolio was two things. First, the permissions model. You're doing permissions properly across all systems. That is genuinely hard and most solutions skip it. Second, the social graph element. It's not just about finding documents. It's about finding people. In our systems there are teams working with Terraform experts that I had no idea existed elsewhere in the organization. Being able to surface that, to connect teams who are working in similar areas without knowing it: that is enormously valuable. Finding information is great. Finding the people who have the knowledge, especially in an organization of hundreds or thousands, is even better.

What's Next for Adaptavist: LLMs, Experimentation, and Staying Curious

GW:  What are you most excited about looking ahead at Adaptavist?

JM:  We are doing a lot of work with LLMs and thinking about how to improve workflows and team velocity using them. We have an experimentation framework and we're building toward flexible integrations with multiple foundation models to see what is best for which use case. A couple of products are incubating in this area that I would love to talk about but our teams would shoot me if I did before they're ready.

More broadly, I see it as very much an explore-and-exploit moment. We are firmly in the explore stage. We need to stay curious and learn what actually changes before we nail things down. The landscape is moving so quickly. What works today may work better tomorrow or may be displaced by something completely different. Being flexible around that is absolutely key. And the thing I always look for in technology: how does it enrich human connection? How does it enrich shared understanding and community? Because that's where all the real power is. The organization is its people and how they connect. Everything else is infrastructure.

GW:  Thanks so much, Jon. I really appreciate you joining us.

JM:  Thanks Gareth. It's been a good conversation. I've enjoyed it.