540 Issues. 745 PRs. One Builder. 60 Days.

In Q1, I started, completed, AND deployed 540 tracked issues for my main client.

Since the first of the year, we've been experimenting together on finding the limits of what one single product builder can do with close to full focus and an army of agents — while still keeping understanding of the system high and "cognitive debt" low.

The answer, at least for me? 745 merged PRs in ~60 working days.

The headline advancement: a full, feature-flagged data platform migration with write-throughs and the works — something which has taken a year, often more, in the 6 or 7 times I've participated in them in my career.

Why migrate the data platform? It had been killing our pace of iteration for the last 18 months I'd been involved.

All that without stopping feature development and improvements.

While this was happening:

• Launched a partner API from zero
• Introduced multiple new features within the main product
• Tightened onboarding + upgrade flows
• Squashed 99 bugs
• Closed 13 critical security vulnerabilities
• 25 UX improvements
• Upgraded years out-of-date front end libraries

The key: not just the agents and the workflow — which are the primary drivers of pace — but also a heavy investment in CI/CD and testing.

I built a full Playwright regression suite from scratch first, as a backstop for all these types of major changes.

There was one incident during all this where I pushed it just a bit too far. Luckily the impact was low — a good reminder that judgement is critical but can always be improved.

This is what's possible now. To go fast, build safety into the workflow.

Join the conversation on LinkedIn