ABOUT THIS EPISODE
Pair programming with AI gets you 20 to 30% faster. Robert Brennan builds for the workflow that gets you 10x.
In this episode of First Commit, Eliya Elon and Laura Hamilton sit down with Robert Brennan, CEO and co-founder of OpenHands — the open source research project that became one of the most starred AI coding agent repos on GitHub. Robert has spent over a decade in commercial open source. Early on, as Cursor blows up and every competitor pivots toward an IDE, his team makes the opposite call: the IDE is an old form factor, and what developers need is an interface for managing fleets of agents. He walks through what that bet looks like two years in, from the banks that refuse to build on closed source to the critic model his team trains to grade its own agents.
Join us as we explore:
- The Dark Factory: What software development looks like when agents write, review, merge, and monitor code end to end — and the two jobs Robert thinks stay human.
- 20% vs. 10X: Why the pair programming model caps out around a 30% gain, and where the order-of-magnitude lift actually sits.
- The Per-Token Trap: Robert on refusing usage-based pricing while competitors post eye-popping revenue at negative gross margin.
- Open Source as Distribution: Why large banks tell OpenHands they will only build on infrastructure they own in the long term.
TRANSCRIPT 🔗
00:00:00.239 --> 00:00:13.119
The most common agent like multi-agent setup that I see today is I'll have one agent that I control doing the implementation of a particular feature or fixing a bug or something like that.
00:00:13.279 --> 00:00:17.039
And then another agent that my team controls does the code review on that.
00:00:17.280 --> 00:00:21.839
And like, you know, my team controls the prompts for the code review, I control the prompt that drives the implementation.
00:00:22.320 -->…