The WisdomBridge assurance flow map
From the first AI TrustGap to continuous assurance: how an engagement moves, what each stage produces, and where independence is protected.
- 01AI TrustGapSelf-attestedPosition · coverage · risk tier
- 02ScopingAgreed in writingScope · access · exit criteria · price
- 03Assurance auditControls & documentationControl findings, owners
- 04Independent evaluationBehaviour under testScorecard · issue log · release gate
- 05Remediation supportSeparate teamFixes implemented
- 06Re-evaluationSame suitesDelta: fixed · new · regressed
- 07Continuous assuranceKeeps pace with changeDated scorecards over time
Assurance is not a single test. It is a sequence, and each stage produces something the next one depends on. The map above shows the full flow. Most organisations enter at the first stage; some join at an audit or an evaluation. Every path ends in the same place — assurance that keeps pace with a system that keeps changing.
1. AI TrustGap — self-attested
A structured assessment that measures the gap between confidence and evidence across up to ten assurance areas. Produces: TrustGap position, evidence coverage, indicative risk tier, findings register.
2. Scoping — agreed in writing
What is in scope — one agent, one AI solution, or a programme of many systems — the claims under test, the access we need, and the exit criteria. Engagement length is set here, from the number and complexity of the systems in scope. Produces: a written scope and a fixed price.
3. Assurance audit — controls and documentation
Policies, ownership, documentation, data handling, supplier management, release and incident procedures — examined against the evidence, not the intent. Produces: control findings with severity and owners.
4. Independent evaluation — behaviour under test
Held-out task suites, an adversarial red-team, robustness testing and, for agents, testing across the full loop: understand, observe, locate, act, verify. Produces: scorecard, issue log, release-gate verdict.
5. Remediation support — a separate team
Help implementing the fixes, delivered by a team that is kept separate from the evaluators. The people who help fix a system never re-evaluate that same system. That safeguard is what keeps the next stage honest.
6. Re-evaluation — the same suites, the delta reported
The original test suites run again. What was fixed, what is new, what regressed. Produces: an updated scorecard and verdict.
7. Continuous assurance — keeping pace with change
Models update, prompts change, tools are added. Scheduled and change-triggered re-evaluation keeps the evidence current. Produces: dated scorecards over time.