Runway: Agents drove 38% of merged PRs in October, median approval wait fell to 5 minutes
runwayml · x · 2026-11-21
Runway engineer Alejandro Alvarez details how the company gradually handed its software delivery pipeline to agents:
- Over the summer, the share of human review time spent on agent-written code in the core API repo more than tripled to about one in five, and PRs awaiting human approval doubled.
- Instead of asking agents for better changes, the team let them take over each stage of delivery (write, CI, review, approve, QA, fix) and defined acceptance criteria for each handoff in advance.
- Results: cloud agents drove 38% of merged PRs in October, up from 0.1% in June; median wait for first approval fell from over two hours to about five minutes between September and mid-November; weekly merged PRs more than doubled.
- Agent-approved changes were reverted no more often than human-approved ones.
The system reviews changes and tests them in a browser, reserving humans for high-risk decisions. Inspiration came from Anthropic engineer Felix Rieseberg's "How to be more ambitious"; OpenAI has described a similar shift building a product from an empty repo.
More from coding & agent
- Haijun Xia pushes back on 'UI is dead': chat is a UI too, and far from sufficient — sh_reya · 2026-11-21
- Vespa co-founders launch Hornet: solving retrieval for the agentic era — jobergum · 2026-11-21
- New Company Agent Bench: Gemini Neon leads at 87.4%, simple API harnesses beat products — philhchen · 2026-11-21
- Head-to-head coding test: Tencent Hy4 Flash finishes in 7:38 vs DeepSeek's 90 minutes — HeyToha · 2026-11-21
- Antigravity's new /teamwork feature lets you assign tasks to multiple agents at once — venturetwins · 2026-11-21
- Zed CEO mocks AI PR spam: opening 100 PRs to fix one bug isn't the flex you think — zeeg · 2026-11-21