Run the 30-minute demo

One system, one story. The audience should leave remembering VNT-02 — Vantera's SuccessFactors candidate-screening system — and its journey, not seven dashboards. You present as Marcus Feld, Head of AI Governance at Vantera Industrial. Everything is one estate, one identifier, three regulatory lenses.

Before you start: confirm with your TrustedAIGov contact that the demo-vantera-industrial tenant is seeded in your environment and that you can sign in. The guided teleprompter lives at /demo/guide and is role-aware for this tenant.

The seven acts (≈32 min; trim Act 1 to hold 30)

• Cold open (2 min) — the question. "What AI are we running, who owns it, what does the EU AI Act mean for us? Last year it took six weeks and the answer was wrong twice." Open /app/home (executive posture) then /app/estate (the governed systems).
• Act 1 — Discover (3–4 min). /discovery-center → enterprise intake: the estate arrived from SAP application inventory and Ariba vendor-register exports, normalised into candidates. Beat: an M365 admin export surfaced a shadow tool in the review queue (/discovery-hub). Honesty line: "We don't magically scan your SAP systems — we industrialise what your teams already export."
• Act 2 — Understand (4 min). Open the flagship AI System Record at /workspace/system/{VNT-02}. Business process, named human owner, lifecycle, and the AI BOM component graph: SuccessFactors module → SAP-operated ranking model, supplied by SAP SE. Point at the identifier: "remember this."
• Act 3 — Govern (4 min). Same record, governance layer: risk assessment, derived controls, evidence with SHA-256 + expiry + review state. Show one green chain end to end: control → current evidence → integrity verify.
• Act 4 — ISO 42001 (5 min). /app/aims/journey: scope (works council as an interested party), the same estate and same identifier, SoA with a real exclusion justification, impact assessment for VNT-02 with independent approval, competence, internal audit with findings, CAPA. "Nothing was re-entered — the AIMS reads the same records."
• Act 5 — EU AI Act (5 min). /app/euaiact/journey for VNT-02: role = deployer; the prohibited add-on screened out under Article 5; Annex III employment category as prospective high-risk (obligations bind 2 December 2027 — say the date precisely); the voluntary FRIA (/app/euaiact/fria); and Article 50 transparency for Joule and the support chatbot (/app/aims/transparency) — a duty that is live today.
• Act 6 — Something goes wrong (4 min). /app/aims/serious-incidents: the support chatbot gave wrong safety guidance → incident record, awareness date, the derived statutory clock, and the assessment that it is not Article 73-reportable (not high-risk; Annex III not yet in force). Then the CAPA, the change-control record (/app/aims/change-control), and a planted control weakness on VNT-03 caught by the evidence-expiry engine, sitting in /workspace/queue.
• Act 7 — Assure (4 min, close). Management review on a frozen snapshot; the auditor's PBC workspace; /app/deliverables → download the signed dossier ZIP and open manifest.json to show the hash chain. Finish on /app/euaiact/posture. Last line: "Six weeks and wrong twice — versus this. Same system, same identifier, three regulatory lenses."

Access

The demo runs in the customer application (not this Learn site): app.trustedaigov.io. Every act degrades gracefully to its JSON API if a page misbehaves — a useful fallback, not a talking point.

Q&A traps to prepare: SAP Agent Hub — "a complement; we'd ingest its inventory, it has no ISO/Act workflows." Is this SAP-certified? — "not required for this architecture; nothing touches your SAP systems." The full act-by-act narration is in the /demo/guide teleprompter.