Everyrequirement,proved in theclient's org.
Delivered is notthe same as proved.
One requirement is one node.Proof is reading every node back.
A requirement card pulls back to reveal that it is one node in a large lattice of requirements, none of which is currently proved.
The loop
How it works.
Spec in
Paste the client's spec, SOW or RFP. The platform extracts it into requirement cards you can actually read.
Scope review
Edit, reject, approve. The approved scope is what gets built — nothing else.
Connect their org
OAuth into the client's own Salesforce sandbox or Marketing Cloud tenant. Their org, their credentials.
Dispatch
Capabilities write real metadata — objects, fields, flows, permission sets, journeys, data extensions.
Verify & sign
An independent verifier reads the org back and checks the value, not just that something exists. What it can't prove stays un-green.
Proof, not assertion.
Green is earned, not stamped.
A requirement turns verified only when a verifier reads the client's org and the value matches the spec. Existence is not proof.
See howIt writes into the real org.
Not a simulator, not a mock. Capabilities dispatch against the client's own sandbox or tenant.
See howOne loop, two clouds.
Salesforce core and Marketing Cloud run through the same spec-in loop, the same verifier, the same evidence pack.
See how
What it can write.What it can prove.
Record types and business processes, provably deployed
Case layouts, assignment rules, and the permissions to see them
Journeys, data extensions, and the audience they actually target
Reads the org back and compares the value, not the existence
A signed pack your client can hold you to
Sales Cloud: Record types and business processes, provably deployed
From our own
regression runs
Pick a requirement.See what the org said.
The import definition exists in the tenant and was found by key —the run reused it instead of creating a second one.destinationName = NT_Import_Staging_Contacts, updateTypeId = 0,allowErrors = false, fileType = CSV.Read back by a separate script that trusts nothing the dispatch reported.
These are runs against our own regression project, adjudicated by an independent verifier and, for the two Salesforce items, re-checked by an adversarial reviewer. We have no customer engagements to show yet.
What the platform measures.
- Verifier-signed runs365Dispatch runs an independent verifier read back and signed off.Source: run ledger
- Verifier refusals415Runs the verifier would not sign. They stayed un-green.Source: run ledger
- No verdict recorded440Runs carrying no verifier verdict at all. Disclosed, not hidden.Source: run ledger
- Capabilities registered58Distinct capabilities that can write metadata into a connected org.Source: capability registry
- Clouds addressed2Salesforce core and Marketing Cloud, through one loop and one verifier.Source: capability registry
Counted 26 July 2026 over 1,220 dispatch runs. These count our own system objects and our own runs against our own connected tenants. They are not customers, not deliveries, and not outcomes.