Home  /  Design partners

Bring us one AI decision.

Show us one consequential machine decision. Dynamic Feed will map the evidence chain, identify what can be verified, and demonstrate the receipt that remains afterward.

Who this is for.

Teams whose software or AI agents make decisions that someone later has to defend: an approval that moved money, an alert that was escalated or dismissed, a route that was changed because of weather, a vendor that was cleared. If a regulator, carrier, court, customer or your own postmortem could one day ask "what did the system know when it acted?" — that decision is a candidate.

What a Decision Receipt proves — and does not

A Decision Receipt is a signed, independently re-verifiable record of the exact inputs a system saw, when it saw them, and the verdict it reached. It is tamper-evident: any later change to the record breaks the signature. It is evidence, not certification — it does not by itself prove the decision was correct, safe or complete, and it never claims events that were not observed. That boundary is stated on every receipt.

A bounded 30-day pilot.

Scope

One workflow only

We instrument exactly one decision path end to end. No platform migration, no long-term commitment during the evaluation — the pilot exists so both sides can see the receipt that remains.

Your part

Expected participation

One technical contact, a description of the decision and its data sources, and a REST or MCP integration point we can call or that calls us. Typical engineering effort is hours, not weeks.

Our part

Evidence-chain mapping

We map which inputs feed the decision, identify what can be cryptographically verified today, wire signed evidence for those inputs, and hand you the verification path — the same one on /verify.

Exit

Measurable success criteria

Agreed up front: for each covered decision, a receipt exists, verifies in your own browser or CI, and honestly records freshness and source state — including stale and unavailable moments. If that is not met, you have lost thirty days, not a roadmap.

Workflows this fits.

Agents

AI-agent action approval

An agent is about to act with real consequences. The receipt preserves what it knew — data, freshness, checks — at the moment of approval.

Security

Vulnerability escalation

A CVE is escalated or deferred. The receipt records the exploited-in-the-wild evidence and advisory state behind the call.

Procurement

Vendor due diligence

A vendor is cleared. The receipt preserves the registries, sanctions and identifier checks consulted, with timestamps.

Logistics

Weather-dependent decisions

A route, sailing or spray window is approved against weather data. The receipt shows exactly which observations backed it.

Operations

Alert verification

An operational alert is confirmed or dismissed. The receipt preserves the corroborating sources and their agreement or divergence.

Everything else

Your consequential decision

If it consumes external data and someone must defend it later, describe it below — mapping it is exactly the pilot's first step.

Start the conversation.

Async only — describe the decision, we reply by email with the evidence-chain map. No calls required, ever.

Please add your name.

A valid work email is required — it is how we reply.

Please describe the decision — one or two sentences is enough.

Privacy: this form sends only what you type above to our enquiry store (rate-limited, access-restricted) and notifies the founder. Please do not include passwords, API keys, production credentials, customer data, confidential evidence, full prompts or raw decision logs — the pilot never needs them. Analytics on this page record event names only, never form contents.

Verify a live receipt yourself on /verify · REST documentation · MCP connection · JavaScript & Python integration · public fixture repository · what a Decision Receipt is.