Blackboard

Shared state board with a coordinator picking the next contributor each round. Classical AI: Erman et al. 1980 (Hearsay-II). Han & Zhang 2025 revived for LLM agents (arXiv:2507.01701). Salemi et al. 2026 (arXiv:2510.01285) reports +13–57% relative improvement on data-discovery tasks.


                ┌───────────── shared blackboard ─────────────┐
                │   facts · hypotheses · partial results       │
                └────▲──────▲──────▲──────▲────────────▲───────┘
                     │ r/w  │ r/w  │ r/w  │ r/w        │
                     │      │      │      │            │
   prompt ──► coordinator ──► picks who acts next      │
                     │      │      │      │            │
                  agent A  agent B  agent C            │
                     │      │      │      │            │
                     ▼      ▼      ▼      ▼            │
                              decider ◄────────────────┘
                                 │
                                 ├─ done? ──► output
                                 │
                                 └─ not done ──► next round

Pattern

Initialize the blackboard with the user’s problem statement.
Each round, the coordinator reads the blackboard and picks one agent to contribute next (or terminates).
The chosen agent runs with the blackboard view in its prompt and produces a contribution.
The contribution is appended to the blackboard.
Loop until the coordinator terminates or max_rounds is hit.
The decider synthesizes the final answer from the blackboard. decider=None falls back to the last answer-kind contribution, or the most recent contribution if no answer-kind exists.

Usage


from loomflow import Agent
from loomflow.team import Team
 
hypothesis = Agent("Propose hypotheses about the data.", model="claude-opus-4-7")
evidence   = Agent("Search for evidence and add it to the board.",
                   model="claude-opus-4-7", tools=[...])
critic     = Agent("Critique the current top hypothesis.", model="gpt-4o")
 
coordinator = Agent(
    "Read the blackboard. Decide which agent contributes next, or "
    "terminate when the problem is solved.",
    model="claude-opus-4-7",
)
decider = Agent(
    "Read the full blackboard and produce the final answer.",
    model="claude-opus-4-7",
)
 
team = Team.blackboard(
    agents={"hypothesis": hypothesis, "evidence": evidence, "critic": critic},
    coordinator=coordinator,
    decider=decider,
    model="claude-opus-4-7",
)
 
result = await team.run("Investigate why retention dropped in March.")

When Blackboard pays off

Data-discovery / investigation. Agents contribute partial findings that compound on the board.
Hypothesis generation + verification. Propose / search / critique cycles.
Loose coordination. When neither hierarchical (Supervisor) nor peer-handoff (Swarm) fits.

Cost: 3–5× single-agent. Reserve for tasks where the shared workspace genuinely accelerates over hierarchical delegation.

Coordinator API. The coordinator is itself an Agent that returns one of: a worker name to call next, or a terminate signal. The framework parses the coordinator’s structured output. See the docstring on BlackboardArchitecture for the exact protocol.