Multi-Agent Team All architectures
v6 · architecture

Blackboard

agents share one workspace; a controller picks who acts

Blackboard architecture

v6 blackboard

Agents share a structured blackboard. A controller inspects the board each round and selects who acts next based on its content — not a fixed schedule. The chosen agent reads the whole board, contributes a section, and writes back, until a solution emerges. Best when an answer assembles from many partial contributions.

Analyst
Planner
Critic

agents share one workspace; a controller picks who acts

How it works
  1. 1Agents share a structured blackboard of named sections.
  2. 2Each round, a controller inspects the board and picks who acts next based on its content.
  3. 3The chosen agent reads the whole board, contributes a section, and writes back.
  4. 4The loop continues until a solution emerges or the board stops changing.
When to use

Problems whose answer assembles from many partial contributions converging on a shared artifact.

Trade-off

Controller selection can loop; no direct peer messaging means coordination is slower.

Agents
AnalystFrames the problem and facts
PlannerProposes the approach/solution
CriticStress-tests and refines the board
Note

Content-driven control (who acts next depends on the board state) is the differentiator. It’s powerful for shared-intelligence problems but slower than direct messaging.