Skip to content

0002 dual model ai consensus

github-actions[bot] edited this page Jul 8, 2026 · 1 revision

ADR 0002: Dual-Model AI Consensus (Gemini + Claude)

Status

Accepted

Context

The protocol needs to autonomously judge whether a freelance deliverable meets acceptance criteria, without a human in the loop for the common case. A single-model verification pipeline was the initial design.

Decision

Require independent, parallel evaluation from two models with no shared context (Gemini and Claude), and only execute an autonomous settlement when both agree above a confidence threshold. Disagreement or sub-threshold confidence escalates to human arbitration rather than picking a side.

Rationale

A single model create a single point of failure — both technically (one exploitable prompt-injection vector) and epistemically (one model's systematic bias or blind spot decides real fund movement unilaterally). Two independent models raise the bar for manipulation and reduce the odds that a single hallucination triggers an incorrect on-chain settlement.

Consequences

  • Roughly 2x the AI API cost per verification compared to single-model.
  • Adds latency (parallel calls still bound by the slower of the two responses).
  • Introduces a meaningful ESCALATE path — the protocol must have a functioning human arbitration flow, not just an automated one, or a nontrivial fraction of jobs will stall. This is currently the oracle authority keypair signing manually; multisig arbitration is tracked as pre-mainnet work.
  • The threshold values (0.80 approve / 0.75 reject) are currently static constants, not on-chain governable — a future version may need these configurable per-job or per-category of work.

Clone this wiki locally