Give us one consequential AI workflow. In 10 business days we put deterministic controls in front of it, intentionally exercise prohibited behavior, and deliver signed, offline-verifiable decision evidence in an examiner-ready package.
$12,500 · 10 business days · 50% before kickoff
100% credited to a design partner if signed within 90 days.
Model-risk teams can explain a model. What they cannot usually produce, months later, is the policy version that was in force at the moment of a specific decision — in a form someone outside the team can verify. That is what an examiner will ask for, and it is the one artifact most AI deployments never built.
The sprint builds it. Deterministic controls sit in front of your workflow; prohibited behavior is exercised on purpose; every allow and block is recorded and signed. You walk away with evidence you verify yourself, without calling us.
CoreGuard in front of the nominated workflow, in a non-production or shadow environment you approve.
ALLOW / BLOCK / human-hold rules for that workflow. No LLM in any verdict.
Including intentional prohibited-behavior exercises — so a block is demonstrated, not claimed.
Recorded ALLOW / BLOCK decisions from the sprint window, each signed (KMS ECDSA P-384).
The full package also includes the offline-verification bundle, an architecture diagram of where the gate sits relative to your stack, findings, remediation recommendations, an executive summary, a governance brief, and a production roadmap from sprint → design partner → production CoreGuard.
| Step | What happens |
|---|---|
| Scoping call | You pick the workflow; we agree the environment and the prohibited behaviors to exercise. |
| Control + policy | CoreGuard is configured against your written rules for that workflow. |
| Exercise | We run the workflow against the controls — including the things that should be blocked. |
| Evidence + walkthrough | Signed records, offline verification, findings, and the examiner-ready package. |
The evidence is not a PDF you take on faith. Every certificate is signed and carries a content hash. Your team checks it against our published key at evecore.com/.well-known/eve-pubkey — no API call, no account, and no dependency on EVE still existing in five years. That is the difference between evidence and a claim.
Stated plainly, because the boundary is the point of the product.
| Evidence Sprint | |
|---|---|
| Fee | US $12,500, fixed, for one consequential workflow |
| Payment | 50% wired before kickoff, 50% on delivery |
| Duration | 10 business days |
| Credit | 100% credited to a design partner if the SOW is signed within 90 days of sprint end |
| Paperwork | A design-partner SOW or a one-page engagement letter — not an MSA chain |
The package is yours either way. There is no obligation to convert, and the evidence you produce stays useful even if you never buy another thing from us.
A 15-minute scoping call to nominate the workflow and agree the environment. First half wired before kickoff; the examiner-ready package lands in ten business days.