Vermilion Cliffs Notes: From MVPs to Agents, How Freeplay Is Helping Teams Ship AI Products
Eval loops and why your best AI workflows might look more like a Slack thread than a sci-fi movie.
In our inaugural episode of Vermilion Cliffs Notes, I sat down with Ian Cairns, CEO and co-founder of Freeplay, to talk about what it really takes to build and ship AI-powered products today. Ian’s background isn’t your typical ML founder story. He’s a product guy (with an English degree, no less) who’s spent his whole career working on developer products, including as Head of Product for Twitter’s developer platform. He got swept up in the generative AI wave not from a research lab, but from curiosity, momentum, and the itch to build. Freeplay also just launched to general availability alongside $5.6million in new funding.
Ian and his team have quietly become a go-to platform for engineering teams weaving LLMs into their apps. But what makes Freeplay different is how deeply they’ve thought about true production workflows: They’re not building a toybox for prompt engineers. They’re building infrastructure for teams that need to debug, evaluate, and ship AI features in production. And they’re doing it with some dry wit and a serious bias for practicality.
In our conversation, Ian talks about how they got started, what they learned from their first 75 customer interviews, and how the teams that are moving fastest today are the ones that’ve figured out how to put non-engineers in the AI loop without breaking everything.
Catch the highlights below and watch the full interview here:
“We were just purely pumped about the tech”
Before founding Freeplay, Ian and co-founder and CTO Eric Ryan worked together for a decade, first at Gnip and then Twitter. In late summer 2022 they started tinkering with OpenAI’s GPT-3 model via the APIon nights and weekends. Ian, an English major who got particularly inspired by “English as a programming language,” saw the potential for something different:
“This was the first time in a while we were purely pumped about new technology… Neither of us went full crypto, but this? This felt like it was going to change the world.”
A good reminder: Sometimes the best companies don’t start with a master plan. They start with curiosity and the right person to build alongside.
“We talked to 75 people in our first 90 days”
Instead of hiding in stealth, Ian and Eric went full pavement-pound mode, talking to anyone who would take the call. That early feedback loop shaped Freeplay from day one:
“We didn’t want to just be a future-looking lab. We wanted to help teams solve real problems today. So we got out there. We talked to 75 people in the first 90 days.”
It paid off. Within months, they had their first pilot customers, and a clear path to building something people actually wanted.
“The eval cycle that took a month now happens in an afternoon”
One of the most compelling parts of the conversation was how Ian described collaboration inside AI teams using Freeplay. Instead of AI being the engineering team’s problem, domain experts, from customer success to finance, are increasingly in the loop:
“We’re seeing workflows where the engineering team wires it up, then moves on. And the domain experts, the people who actually know the context, take over evals, QA, and prompt iteration. The cycle that used to take a month? Now it happens in an afternoon.”
This is what "cross-functional" actually looks like in AI right now: teams where everyone contributes to product quality, not just shipping code.
Final Thoughts
Ian’s story is one many founders will recognize: a technical product mind who finally found the tech that made him want to go all in. Freeplay grew quickly by staying close to its customers, building for real-world AI use cases, and making hard things feel more usable.
Whether you're just starting to build with LLMs or scaling a full agent workflow, this episode is worth a listen. Full interview here.