Field evidence

Measure before, during, and after AI enters the workflow.

A field study needs a baseline, fixed method version, real follow-up outcome, privacy review, limitations, and independently verifiable participation.

0 publishable studies

No field study has passed the evidence gate.

AI Amigos will not convert planned pilots, private conversations, generated examples, or protocol demonstrations into public impact claims.

Inspect the methods Propose a governed pilot

Required field-study record

1. Pre-registration

Job, baseline, sample, metric, owner, success criterion, stop criterion, method version.

2. Comparable execution

Fixed cases, system versions, cost, latency, failures, human overrides, deviations.

3. Follow-up outcome

Observed result after use, date window, confounders, limitations, reversals.

4. Verification

Consent, privacy decision, reviewer scope, evidence hash, outside corroboration.