In this edition of Luminar’s Founder Series, we spoke with Adam Tornhill, founder of CodeScene, about the path from software engineering and psychology to building one of the most thoughtful platforms in developer tooling.
Adam’s background is unusually broad: nearly four decades of programming experience, a master’s degree in psychology, and a deep belief that the best software products are built by understanding both code and the people behind it. In our conversation, we discussed the origins of CodeScene, the company’s research-driven approach, and why he believes agentic coding will make engineering more important, not less.
You have a background in both software engineering and psychology. How did that combination come about?
I’ve been writing software for a long time, next year I’ll celebrate 40 years as a programmer. After about ten years of large-scale software development, I realized that many of the biggest problems in software weren’t technical. They were about collaboration, communication, and how people solve problems together.
That led me to study psychology. I spent six years earning my master’s degree while working full-time as a consultant, and that combination has been incredibly useful, especially now in the age of agents.
Was there a practical advantage to mixing the two fields?
Absolutely. The code health metric we developed at CodeScene has its roots in cognitive psychology. We wanted to measure code in a way that reflects how it actually affects the humans working with it.
That’s why our model looks at around 25 factors tied to how hard code is on the brain. It’s been important for connecting code health with real engineering outcomes like shorter lead times and fewer defects.
How did your book lead to CodeScene?
I was working as a consultant and often had to tell the business that a feature couldn’t be built on the timeline they wanted because of technical debt or poor code quality. But I didn’t have a convincing way to prove it.
So while studying psychology, I started looking at ideas from criminal psychology and applying them to software. Instead of tracking offenders, I was finding bottlenecks and risky areas in code. That work eventually became my book, Your Code as a Crime Scene, which later became the foundation for CodeScene.
Did it feel connected in hindsight?
Yes, definitely. I had a clear vision for what I wanted CodeScene to become. I was just overly optimistic about how quickly I’d get there - which, I think, is a classic founder mistake.
What surprised you most about becoming a founder?
In the early days, you do everything yourself. The upside is that you can’t blame anyone else if something fails - and you learn a huge amount very quickly.
I’d never done sales before starting CodeScene, but talking to customers turned out to be one of the most valuable things I ever did. It made me a better product builder.
What also surprised me was how long enterprise sales cycles can be. A company may say they want to try the product, but in reality it can take months. Today, the bar is even higher because security and trust are non-negotiable.
What advice would you give first-time founders?
Be ready to do everything yourself in the beginning, and don’t underestimate patience. Building a company almost always takes longer than you think.
What CodeScene does
How would you explain CodeScene to someone unfamiliar with it?
CodeScene does three things.
First, it gives you a web-based visualization of your source code so you can identify the riskiest parts of the codebase and understand what they mean in terms of development speed, defect risk, and AI-readiness
Second, developers can plug CodeScene into their IDEs or MCP server to get those insights and AI-safeguards directly in their workflow.
Third, it guards any changed code through automated pull request reviews. We even have agents that can uplift any unhealthy code.. In other words, it supports the full software development lifecycle.
How has the agentic revolution changed how you work?
For me personally, it’s been the best thing that’s happened in decades. In the early years, I wrote the first version of CodeScene myself. But over time, as the company grew, I had less time to contribute hands-on.
Now I can spend 20 to 30 percent of my week on hands-on work and still have a real impact. If you have domain expertise and a clear idea of what you want to build, agents can help you move much faster.
You’ve worked with Luminar for several years now. What changed when the company became venture-backed?
It was a massive shift. Before that, CodeScene was bootstrapped and quite small, with a strong R&D focus and very little sales or marketing budget.
Once we became venture-backed, we could broaden the team and invest more in the business. I also had to change as a founder. Before, I could keep everything in my head. Once the company started moving in parallel across many fronts, I had to let go of some control and trust the team more.
That was a major leadership lesson for me. You don’t want to be the bottleneck.
How do you see software development changing over the next few years?
I don’t think the future will look radically different in broad terms, but I do think teams will become smaller and take on broader responsibilities. More tasks will be automated, and people will need to combine more roles.
At the same time, companies that rush into agentic coding without the right engineering foundation risk quality issues. That’s why we’re focused on helping teams adopt these methods safely and effectively.
What’s CodeScene’s longer-term vision?
We want to make it easier than ever to adopt agentic coding while keeping quality under control. That means supporting developers, coding agents, and leadership teams with the right insights at the right level.
Software is a strategic asset, and companies need to know whether AI adoption is actually helping. That’s the problem we’re building toward.
What helps you stay grounded as a founder?
I still like contributing code and features myself, because that’s how I learn. I also stay close to customers. Even with a strong sales team, I like being involved with our accounts because the feedback is so valuable.
I’ve also learned the importance of stepping back. In the early years, I was working 70 to 80-hour weeks, which isn’t sustainable. Now I try to take a walk every lunch break. That’s often when the best ideas come.
Thanks to Adam for the thoughtful conversation. CodeScene is helping teams build better software with stronger code health and smarter agentic workflows, and we’re excited to follow what comes next. Stay tuned for more Founder Series insights.