Holistic AI

Are scope, spend, and authority enforced while the agent runs?

Re-grounded 2026-10-04 against the current capture (2026-09-23). Vendor describes deployment gates, approval sign-offs and remediation with human oversight, and says Guardian Agents provide run-time oversight with Operative Agents able to block, redirect, stop or escalate an action. Not 3: scope, spend and authority limits are not published as enforced at run time. The prior rationale cited "kill switches" from the 2026-07-31 capture; that term no longer appears on the page, replaced by the intervention wording above.

Our automated recount on Sep 27, 2026 counted enough evidenced fields for 1/3. It never reads what those fields say, so it does not overturn the score above, judged on Oct 4, 2026 — but where the two disagree it is a reason to read this dimension again. Why there are two scores

Is there a tamper-evident record of what it actually did?

Re-grounded 2026-10-04 against the current capture (2026-09-23). A record exists and is customer-reviewable: the page states a full audit trail for regulators and internal review, and "Build audit-ready evidence. Record decisions, versions and mitigations. Export compliance reports for review." Not 3: nothing published claims the trail is tamper-evident or hash-chained. The prior rationale quoted "continuous audit trails, evidence collection and compliance reporting, audit-ready from day one" from the 2026-07-31 capture; that wording is no longer on the page, though the substance survives in the current one.

Our automated recount on Sep 27, 2026 counted enough evidenced fields for 1/3. It never reads what those fields say, so it does not overturn the score above, judged on Oct 4, 2026 — but where the two disagree it is a reason to read this dimension again. Why there are two scores

Are compliance obligations attached per engagement?

Re-grounded 2026-10-04 against the current capture (2026-09-23). Vendor maps controls to the EU AI Act, NIST AI RMF, ISO/IEC 42001 and NYC Local Law 144 and publishes 40+ specialised tests. Not 3: no certification of Holistic AI itself, and no per-engagement compliance schedule, is published on the page checked. The prior rationale cited "automated control mapping and gap analysis" from the 2026-07-31 capture; "gap analysis" no longer appears on the page, control mapping still does.

Our automated recount on Sep 27, 2026 counted enough evidenced fields for 1/3. It never reads what those fields say, so it does not overturn the score above, judged on Oct 4, 2026 — but where the two disagree it is a reason to read this dimension again. Why there are two scores

This score counts Compliance certifications as undisclosed, and that absence is flagged for review — the cited source may address it. If so, this score is too low. How we check absences

Does payment depend on a verified result?

No public evidence found. The page checked states no outcome-linked payment, fee-at-risk or settlement term.

Can the deployment be insured, and is the customer indemnified?

No public evidence found. The page checked names no carrier, cover, limit or customer indemnity.