Stanford paper: one agent per user underperforms a single shared agent on common resources
rohanpaul_ai · x · 2026-10-07
A new Stanford paper finds that when each user's agent acts alone on a shared resource, the group does worse than a single agent serving everyone, across 5 frontier models.
- Shared budgets and calendars are handled better by one coordinator agent than a team of per-user agents.
- Each agent looks sensible individually, but together they overwrite each other's actions and stall as the team grows; with no communication channel, teams collapsed outright in 2 environments.
- On a contested token budget, Opus 5 teams captured 30% of achievable value vs 64% for a single coordinating agent.
- Agents fabricated facts about other users in over half of Claude team episodes.
Practical takeaway: prefer one agent holding everyone's constraints; if running one per user, make reading peers a precondition before committing.
Related event: Stanford Study Finds Multi-User Agent Teams Underperform a Single Agent(2 posts)→
More from coding & agent
- Brett Adcock shows off Handoff, a computer-use agent for shopping and travel — adcock_brett · 2026-10-07
- Decisions API is out; users pitch baking it into ChatGPT for real-time gameplay — flowersslop · 2026-10-07
- Feeding years of WhatsApp and Instagram data to a personal AI agent — we93 · 2026-10-07
- AgentMail launches AgentID, a 'Sign in with Google' for AI agents — ycombinator · 2026-10-07
- Gallery of 226 Claude Opus motion graphics with the prompts behind them — dotey · 2026-10-07
- Open-Source Project With 2,000+ Stars Adds 9 AI Animation Explainer Styles and Full Workflow — dotey · 2026-10-07