01 / Operations lead
The team sets the brief.
Prepare a supplier recommendation using approved sources. Check the required documents. Do not place an order or accept commercial terms. The team delegates the research, not the final commitment.
Examples / an illustrative supplier assessment
Follow one example from a clear brief to a decision someone can explain: what the agent could do, what stopped the workflow and who made the call.
01 / Operations lead
An agent is asked to assess a supplier using approved sources. It may gather evidence and make a recommendation. It may not place an order.
The application enforces the brief. A receipt does not enforce it by itself.
SUPPLIER ASSESSMENT
Scope defined
Chapter 1 of 6Focus this story and use ← → to explore
Product illustration, not a live customer workflow. The full readable account and all alternative outcomes remain below.
The example
The agent does the groundwork. The application holds the limit. A person makes the consequential call. The record keeps the story together.
01 / Operations lead
Prepare a supplier recommendation using approved sources. Check the required documents. Do not place an order or accept commercial terms. The team delegates the research, not the final commitment.
02 / Research agent
The agent gathers the permitted documents, compares them with the checklist and drafts a recommendation. The useful work moves forward without a person reviewing every search or extraction.
03 / Workflow rule
A required insurance certificate has expired. The recommendation may be ready, but the evidence is not. In this scenario, the application routes the exception to a person and holds acceptance.
04 / Authorised reviewer
The reviewer sees the relevant document, the rule it failed and the agent’s recommendation. They can request an update, accept suitable evidence or reject the recommendation. Silence is not approval.
One job. More than one honest ending.
These are alternative paths through the same proposed workflow. None is a live action or a universal verification result.
An updated certificate is received and assessed. The reviewer accepts the recommendation against that version. The original exception remains in the history.
The reviewer requests a current certificate. The job remains unresolved until the new material is assessed. If the review deadline expires, the application keeps acceptance on hold.
The evidence does not meet the organisation’s criteria. The reviewer rejects the recommendation and records why. Any later reassessment is a new, linked decision.
The evidence changes
A legitimate update and a mismatched disclosure need different handling. Neither should quietly inherit the old decision.
Retain the old record. Create a new evidence commitment and link the reassessment to the earlier work. The reviewer makes a fresh decision against the new version.
The integrity check fails. Obtain the original version or record a new one explicitly. Do not display the earlier approval as applying to the changed file.
Report that the evidence is unavailable. The anchor may still exist, but it cannot recreate the file or turn an incomplete check into success.
After the decision
The proposed review package links the brief, evidence and decision so an authorised recipient can inspect the story later.
Meet the receipt tooling →Illustrative package / accepted path
This human-readable package is a product concept, not a literal receipt schema.
Private work. Shared confidence.
In this proposed workflow, the organisation keeps the source material. A reviewer receives selected evidence. The shared anchor records a commitment—not a public copy of the job.
The custodian manages access, retention and recovery.
Disclosure is an access-controlled copy of selected evidence, not unrestricted access to the vault.
A commitment binds to recorded data. It does not reveal or recover that data by itself.
Privacy is not automatic: transaction metadata may be visible, and recipients can retain disclosed material. Evidence protection, careful commitments and retention policies still matter.
Understand the verification boundaries →Why the record matters
The intended benefit is a decision history another team can inspect without rebuilding the job from scattered logs. Pilot work still needs to establish the time saved and the value to buyers.
Matching evidence does not establish its truth. A recorded approval does not make judgement infallible. The useful question is always: what was checked, by whom, and against which version?
See how the product fits →Beyond this workflow
The same need for accountable history reaches into cross-organisation services and physical machines. Each new setting needs its own evidence and assurance.