I've always found joy in learning. Teaching fascinated me too, but I never quite had the patience for it. Or maybe I just never had the chance to learn how to teach.
So for most of my life, I've been a student. Specifically, a student of software engineering.
Last year, while writing a lot of software, something clicked. Software engineering was changing. AI wasn't just changing what we were building. It was changing what was possible to build at all.
I realized the only way I could learn the true potential of this technology was if I challenged myself to build things which were unthinkable before.
So I decided to start something I've always wanted to do. For years I'd wanted to build tools for learning, and now it finally felt possible. Before jumping in, I wanted to know three things:
Could I build a really complex platform with a tiny team, without cutting corners on quality? Could I build something that helps people learn and think better, the kind of problem that actually needs intelligence to solve? And could I have fun while doing it?
The first two turned out to be more possible than I expected. As a seasoned engineer, coding agents gave me a lot of leverage. And even though I'm no machine learning researcher, I could build intelligent systems on top of language models without training anything myself.
The third one, technology couldn't help with. I love engineering, but doing it alone is lonely. You need someone to challenge you, and to laugh at you now and then. That's when I reconnected with Dan. We'd worked closely at Expedia, and after he left, he was looking to build tools for learning too. We clicked immediately. We both believed AI should make people smarter, not become an incomprehensible alien that does all our work while we understand nothing. And we both cared, maybe a little too much, about craft and quality.
AI is a thinking partner, not an answer engine; the moment it becomes a shortcut around thinking, it's working against us.
That belief turned into a few principles we hold ourselves to. AI is a thinking partner, not an answer engine; the moment it becomes a shortcut around thinking, it's working against us. We care about the few insights that change how people understand a problem, not about capturing everything. And we hold our principles firmly and our implementations lightly. We listen to data, but we still decide, commit, and allow ourselves to disagree and commit.
Then we started building. The first thing we made was PeerNotes, a place where teams work through and remember their decisions. You bring the messy raw material, a rough thought, a Slack thread, a meeting. PeerNotes turns that into a clear decision, and keeps the reasoning, the alternatives, and the open questions behind it as team memory. It began as a tool for individual learning, a loop of reading, distilling and revisiting ideas. Along the way, we realized the hardest thinking doesn't happen alone, it happens between people too. So the loop grew to hold both.
Building PeerNotes properly meant solving a lot of hard, unglamorous problems, like keeping data safe and making sure the AI only ever sees what it should. Instead of solving them once and moving on, we turned them into the Obtainable platform, a foundation that everything we build next can stand on. PeerNotes is the proof that it holds up.
Building it also taught me something about what a small team can actually do now. The agents handle a lot of the execution, but the judgment still has to be yours, and that demands a lot of focused, highly skilled work, because making sure the quality holds up means understanding everything the agent is doing.
Intelligence, it turns out, you can borrow. The hard part is deciding what it's for. In PeerNotes, the AI asks questions instead of handing out answers, and points at the gaps in our reasoning instead of covering them up. That feels like the right job for it.
And yes, I'm enjoying it, mostly because I'm not doing it alone.
So now I know how to build it. What I have no idea about is how to get it to the people who need it. I guess I'm still a student after all, and that's the next thing I have to learn.