AI Trust Snapshot — TrustAssay

Operating model & governance

1. Is there a single, clearly accountable owner for AI/QE quality across your organisation?

No Informally, no clear owner Yes, but only for part of the portfolio Yes, a named owner across the full portfolio

2. Is there a governance forum that reviews AI system quality on a regular cadence?

No Ad hoc, only when issues arise Yes, but infrequently Yes, on a defined regular cadence

Test strategy & coverage

3. Do you test AI outputs for bias across different user segments?

No Ad hoc only Periodically As part of every release gate

4. Do you know how much of your test suite is redundant or overlapping, or do you only track total test case count?

We only track total test case count We suspect overlap exists but haven't measured it We've measured it once or occasionally Yes — tracked continuously as part of coverage reporting

Automation & AI-native capability

5. What share of your test cases are AI-generated or self-healing vs. manually maintained?

None A small pilot Meaningful adoption, not enterprise-wide Enterprise-wide, AI-native

6. Are AI agents integrated into your CI/CD pipeline as gating checks, or run separately?

Not used in the pipeline Run manually, out-of-band Partially integrated as gates Fully integrated as automated gates

Delivery economics

7. How long does a typical AI feature take from code-complete to release?

Weeks About a week A few days Under a day, largely automated

8. Has AI adoption in QE measurably reduced cost or cycle time so far?

No measurable change Some early signal Yes, moderate improvement Yes, significant and tracked

AI assurance & regulatory readiness

9. Do you measure hallucination/error rates against a human-verified ground-truth dataset?

No, we don't measure this We spot-check occasionally We measure it periodically Continuously, with automated gates

10. If a regulator or board member asked "why did the AI make this decision," could you answer within a day?

No Maybe, with significant manual effort Yes, with some effort Yes, instantly via an audit trail

11. Do you have evidence mapped to a framework like the EU AI Act, ISO 42001, or NIST AI RMF?

No Aware of requirements, nothing mapped Partially mapped Fully mapped and audit-ready

Talent & skills

12. What share of your QE staff are cross-skilled in AI/ML testing vs. traditional-only?

None A small handful A meaningful minority Most of the team

13. Is there a defined reskilling path for AI testing skills, or is it ad hoc?

No programme Informal, self-directed Yes, a defined but partial programme Yes, a formal enterprise-wide programme

The AI Trust Snapshot