Our story
From side project to the loop we run production on.
Shipwright wasn't planned as a product. It grew out of two engineers trying to get more out of Claude Code, dogfooded on our own delivery work the entire way. Here's the timeline, honestly.
November 2025
Dan starts running Claude Code in the cloud. Dave starts building the first autonomous workflow plugin.
January 2026
The first real autonomous-loop plugin lands in the App Vitals marketplace, inspired by the open-source Ralph pattern. Auto-approve hooks, task types, plugin discovery — the pieces of a self-driving loop start coming together.
March 2026
We start building Vitals OS — and from day one, we run the build through our own emerging Shipwright workflow. First real product built by the thing we were building.
April 2026
The automation backbone goes in: a dev-task runner, agent cron infrastructure, scheduled review/patch/deploy cycles. This is where "autonomous" stops being aspirational.
May 2026
Metrics wired in end to end. We can finally see what the agents are doing, not just trust that they're doing it.
June 2026
Shipwright graduates out of the marketplace into its own repo. We scrub every internal reference and open it up.
Late June 2026
Public launch.
Today
Two of us, shipping 50+ PRs a day, running production infrastructure for clients almost entirely through the loop we built for ourselves.