How AI-native product teams share one repo: thin CLAUDE.md, MCP connections, 16+ skills
aakashgupta · x · 2026-10-24
Aakash Gupta distills Oji Udezue's work making companies AI-native: the fastest teams give PMs, designers, and engineers one shared repo (a "Team Operating System"), with Hannah Stulberg at DoorDash and Gabor Mayer at Google running versions of it. Four layers matter:
- Thin CLAUDE.md: base models are better now, so cut instruction files from 400+ lines to under 80 — keep only architecture and startup essentials that every chat reads, to avoid wasting context.
- Connections: wire up GitHub, Jira/Linear, Notion/Google Docs, Tableau+Amplitude, Qualtrics+Dovetail, and email/calendar via integrations; teams lacking BI access are handicapping themselves.
- Shared skills: encode common tasks (measuring feature effectiveness, recent user research, OKR status) as slash commands — Oji runs 16, Gupta has 41+.
- Rollout: don't merge everyone's OS into a context-eating monster; build iteratively via PRs from first principles.
The payoff: PMs answer fewer questions, teams self-serve, and PMs get time back for strategy and user learning instead of becoming project managers.
More from coding & agent
- Agent with SSH access successfully upgrades an ancient PHP 5.6 server end-to-end — DanielLockyer · 2026-10-24
- DeepMind researcher demos a playable 3D racing game built right on her desk — thursdai_pod · 2026-10-24
- Cut LLM costs: route requests, disable thinking, cache answers, escalate only on low confidence — colinmcnamara · 2026-10-24
- NVIDIA's OpenShell: Apache-2.0 runtime sandbox for autonomous agent fleets, 18.6k stars — petrusenko_max · 2026-10-24
- Claude Code Routines runs overnight, screens 380 clinics into 30 outreach drafts — Arindam_1729 · 2026-10-24
- Factory and Anthropic share customer story on long-horizon coding agents and 'dark factories' — matanSF · 2026-10-24