WorkOS built a software factory: agents pick tickets autonomously, judged by outcomes not PRs
AI Engineer · youtube · 2026-09-27
WorkOS engineer Ryan Cooke explains their internal "software factory": agent TARS lives in Slack/Linear/GitHub and self-assigns tickets via webhooks; a PM agent drafts product docs to solve the blank-page problem; the Horizon orchestration layer fronts an internal MCP gateway he calls the single best investment for such teams. Key argument: measure outcomes (defect rates, recovery time, adoption), not PRs or lines of AI code. Also covers support triage, custom sandboxes, and a company-wide memory layer.
More from coding & agent
- 100,000 sites scanned for agent readiness, growing 10x in a month — EdenEmarco177 · 2026-09-28
- Vibe coding favors games with addon systems—expect them everywhere — marktenenholtz · 2026-09-28
- AI agents' PR flood pushed this team from GitHub runners to on-prem CI — peterjliu · 2026-09-28
- Coding agents write defensive code with overly broad exception catches, and review agents still LGTM it — srchvrs · 2026-09-28
- Asked Claude for a small scrape, woke up to 'three weeks remaining' at 60 MB/s — generativist · 2026-09-28
- Dad and kids vibe code a Zelda-style open-world game with Opus 5.5 in one day — IanArawjo · 2026-09-28