# PactVerity Flagship Pilot Specification v0.1

Status: open for proposals; no flagship pilot or customer claimed.

## Goal

Prove one bounded AI-agent or API-service outcome in 30 days using real measurements, deterministic checks and a reproduction by a second engineer.

## Suitable pilot

- A real but non-sensitive workflow with a named owner.
- One observable result, such as completion, latency, uptime, freshness or output integrity.
- A disclosed evidence origin and measurement method.
- No wallet custody, investment solicitation or private customer data.

## Milestones

1. Baseline: publish the workflow, criteria, expected input and known limitations.
2. Instrumentation: capture real observations without credentials or personal data.
3. Receipt: create and verify receipts through the public beta API.
4. Reproduction: a second engineer repeats the process from the documentation.
5. Case study: both parties approve a factual report linking evidence and limitations.

## Success criteria

- At least 20 disclosed pilot runs across at least 7 calendar days.
- Complete pass and failure outputs retained.
- Reproduction from a clean environment.
- No discrepancy between the case-study wording and underlying evidence.

Selection, work in progress and a successful first-party demonstration are not verified integrations.