Each ensemble is a pre-configured task type with thermodynamic parameters from the NSED whitepaper. The utility curve U(t) = 1-(1-pg)e-Λ(pv-pg)t - βt² predicts accuracy per round, combined with evaluator convergence for optimal stopping.

Deliberation Rounds = the maximum number of propose-evaluate cycles.
Convergence allows early termination if agents reach agreement before all rounds complete.

After each round, a convergence score (0.0 – 1.0) is computed from how much assessors agree on opinion rankings. A score of 1.0 means perfect agreement; 0.0 means total disagreement.

Convergence Threshold: the minimum convergence score to consider "converged" (e.g., 0.8).
Convergence Rounds: how many consecutive rounds must exceed the threshold before terminating early. This prevents premature termination from a single lucky round.

Example: With threshold=0.8 and convergence_rounds=2, if round 3 scores 0.85 and round 4 scores 0.82, deliberation stops after round 4 even if 5 rounds were configured. If left empty, all deliberation rounds will always run.

User tools allow agents to call external functions during deliberation. Each tool call is forwarded to the dashboard for manual response.
The dm_user tool is a common preset that lets agents send messages directly to the user for clarification.