How It Works

The EvidenceCycle Operating System

EvidenceCycle is a structured, repeatable framework for high-stakes AI decisions. It moves you from hype to evidence-grounded action through four phases and core artifacts.

The Atomic Unit: Evidence-Based Claims

Every decision is built from claims examined through this lens:

1.Claim — What is being asserted?
2.Evidence — What supports it? What's the quality?
3.Limitations — What don't we know? What could go wrong?
4.What would change our mind? — What future evidence would strengthen or weaken this claim?

Four Phases

Phase 01Discover
Required

Map the decision landscape. Identify sources, stakeholder constraints, and evidence gaps.

Artifact

EvidenceVault

Curated, traceable sources with actionable summaries and limitations

Phase 02Design
Optional

Distill decision-critical claims and risks into reusable building blocks.

Artifact

EvidenceCards

Decision-ready claim assessments with evidence quality, limitations, and change conditions

Note: Recommended; optional for lightweight workflows

Phase 03Build
Required

Synthesize cards into decision-ready guidance that accounts for your organizational context.

Artifact

EvidenceAtlas

Complete decision package including embedded EvidenceCycle Briefs (procurement readiness, deployment guidance, governance requirements)

Phase 04Test
Optional

Define what to measure before and after deployment. Establish monitoring and escalation protocols.

Artifact

EvidenceBench

Evaluation and monitoring plan with embedded EvidenceCycle Briefs, metrics, thresholds, drift signals, and governance workflows

Note: Optional extension