E10 · Shared state (blackboard)
Family: Multi-agent
Autonomous agents read from and write to a common store, with no coordinator.
Principle
The agents are autonomous and work on a common persistent store: they read what the others found and write their own results. There is no coordinator; each builds on what accumulates. It is the blackboard of distributed systems.
The problem it solves
A collaborative investigation moves forward through accumulated results; a single coordinator would be a bottleneck and a point of failure.
When to use it
- Collaborative investigation, where agents build on accumulated results.
- You want no single point of failure.
When to avoid it
- Stopping conditions are not strong: reactive loops burn tokens without converging.
- Duplicates and contradictory approaches are unacceptable, or agents write to the same state at the same time (a pitfall cited by Microsoft): prefer E1.
Synonyms by vendor
- Shared state (Anthropic (2026))
- Blackboard (AWS, architecture distribuée classique)
Neighbouring patterns
Sources
- Multi-agent coordination patterns: five approaches and when to use them — Anthropic (C. Phillips), 2026-04-10
- Why Do Multi-Agent LLM Systems Fail? — Cemri et al., arXiv, 2025-03
- AI agent orchestration patterns — Microsoft, Azure Architecture Center, 2026-02-12
- Don’t Build Multi-Agents — Cognition (W. Yan), 2025-06-12
Content reviewed on 2026-10-02