Data requirements
A smaller, documented input makes the service safer and easier to audit.
The normal starting point is metadata-only approval policy and event CSVs. Nothing operational should be sent through the public website.
Accepted profile
Fields the working schema is designed around.
- case token
- version token
- review stage
- reviewer role
- event period
- decision state
- evidence state
- exception code
The field list is a planning profile, not a request to email data. The exact schema and secure exchange are agreed in writing after fit review.
Explicit exclusions
Data and authority we do not need.
No legal or compliance determination, autonomous approval, broad workflow transformation, or claim of regulatory certification.
Credentials, unrestricted system access, and unrequested free text are never part of the published plan.
Data lifecycle
Five visible handling stages.
- 01
Agree
Write down the permitted fields, purpose, owner, transfer, retention, and deletion rule.
- 02
Prepare
The buyer exports and normalizes only the agreed minimum input.
- 03
Validate
Schema failures and unknown values are rejected without inference.
- 04
Deliver
Portable outputs carry definitions, exceptions, and confirmation states.
- 05
Delete
Working data follows the written retention rule; the buyer keeps the final pack.
Field glossary
Shared definitions prevent expensive rework.
- Case
- One buyer-defined approval instance represented by an opaque token.
- Version lineage
- The buyer-declared predecessor and successor links between versions.
- Decision state
- The as-reported approval state, without interpretation.
- Required stage
- A step present in the buyer-authored workflow policy.
- Evidence index
- A content-free map of which required events were reported.
- Exception
- A structural mismatch that needs human review.
A useful first conversation
Check the data boundary before discussing a plan.
Describe it in general terms. We will tell you which plan fits, what we would need, and where we would draw the line.