Sharing agent sessions across a team breaks knowledge silos — and AI can summarize your teammates' workflows
steipete · x · 2026-11-04
patrickerichsen describes moving his team to multiplayer agent sessions where everyone can read each other's prompts. Initial fear of exposing 'dumb' prompts gave way to two lessons: everyone talks loosely to agents, and teammates' workflows are a goldmine. He had an agent read his UI/UX colleague's eight sessions and extract the prototyping loop: use the agent as a live prototyping partner while keeping art direction and acceptance decisions yourself, iterate low-fidelity near the real code, and only then push to production. Sharing prompts tears down useless knowledge silos.
More from AGI Musings
- Investor argues scaling laws are correct — it's how we measure data, params and compute that's wrong — JosephJacks_ · 2026-11-04
- ChatGPT's answer to humanity's most important question: get wiser faster than more powerful — UnderstandingBasic82 · 2026-11-04
- As AI learns skills one by one, do you love the work or just being needed? — nathanbenaich · 2026-11-04
- Pedro Domingos mocks LLM reasoning: still CoT and brute-force compute a decade later — pmddomingos · 2026-11-04
- Your job as a scientist is disproving your own results: baselines, leakage, reward hacking — tw_killian · 2026-11-04
- Superhuman CEO Shishir Mehrotra: the five AI debates every board is having — MatthewBerman · 2026-11-04