Evidence-gate card
Verolane AI
Fictional company, real review pattern.
Verolane AI runs an agentic automation platform whose agents take actions inside a customer's systems with reduced human oversight. A large enterprise wants to buy, and its security and procurement teams have sent an AI-governance questionnaire before they will sign.
The team would rebuild the answer by hand, from Slack threads, tickets, model-change logs and the founders' memory. It takes weeks, reads differently every time, and the evidence is stale by the next questionnaire.
Custara prepared a single record for the agentic workflow, with authority boundary, model change control, oversight settings and incident evidence brought into one structure before the questionnaire response was exported.
GER-00X · FICTIONAL
Scope: enterprise questionnaire · Evidence period: current cycle
Record map
- Model change control, status: present
- Agent action and escalation log, status: present
- Human oversight and approval boundary, status: present
- Incident and complaint log, status: present
- Data-source approvals, status: needs source
- Third-party model terms, status: missing
documented at the time
Featured item (expanded): an agent action event. What was done: an agent executed a scoped change. Under whose authority: within the customer's configured approval boundary. On what basis: the inputs and agent version at the time. How captured: documented at the time, from the orchestration log. Kept intact: record reference retained with dated capture details.
Fictional record. Status reflects record completeness, not governance judgement. Not a score, an assessment or a certification.
Data-source approvals still need source, and third-party model terms are missing. The record shows both plainly instead of hiding them.
Shown, not hidden. No score state.
A controlled questionnaire pack formatted to the enterprise customer's structure, scoped to the evidence they asked for.
The same record carried into the insurance renewal and stayed ready for the next enterprise questionnaire.
Custara prepared and organised the evidence. The customer's reviewers decided whether it was sufficient.
Custara prepared the evidence and organised it the way the reviewer asked. The reviewer decided it was sufficient. Custara did not.
Custara prepares the evidence record. It does not assess, score, classify, certify, rate, advise or recommend.