VERA.AI turns the operational evidence you already produce into a regulator-ready pack — with provenance on every field and a tamper-evident audit trail. We prove it first on the EU AI Act's Annex IV, the mandatory document every high-risk system needs. One engine, already running three frameworks.
Built for compliance officers at regulated organisations
EU AI Act Article 11 requires Annex IV technical documentation before placing any high-risk AI system on the market. The document has 9 mandatory sections covering data governance, risk management, performance metrics, human oversight, and more. Today, organisations assemble it manually in Word and PowerPoint, pulling from a dozen systems. It takes 6-9 months. We do it in days.
Governance platforms like Credo AI now generate policy-level documentation, and labeling tools produce data. But the deep technical evidence regulators examine first — architecture, traceability, provenance — is the cited gap. VERA.AI goes deepest there, affordably, and Italy-first.
Import from Encord, Labelbox, Scale AI for labeling data. Import from Credo AI, OneTrust for governance metadata. Or use our native labeling.
Answer 8 structured questions about your AI system. We classify against Annex III and map to the correct Annex IV obligations.
All 9 sections of Annex IV fill from connected sources: purpose, training data, model architecture, performance, monitoring, oversight.
Generate a signed PDF + machine-readable JSON. Refresh annually or on substantial modification. Audit chain proves integrity.
VERA.AI is a regulatory evidence platform. The same engine — scoring, provenance, tamper-evident audit, multilingual rendering — now powers three frameworks, and adding another is a content-and-rules exercise, not a rebuild.
Technical documentation for high-risk AI systems. The original framework, in full regulator-grade depth.
AI risk-management evidence for organisations aligning to the US framework alongside the EU.
Continuing-compliance evidence for large-investment megaprojects (Ley 27.742). A non-AI regime — proof the platform generalises. Bilingual Spanish/English packs.
You pay a base platform license plus a fee per evidence pack you generate. Annual refresh keeps your packs current as regulations and your AI systems evolve.
Annual contract per organisation. Includes platform access, support, and Annex IV template engine.
One per new high-risk AI system. Then €3,000/year refresh per existing pack.
NIST AI RMF and ISO 42001 evidence packs for organisations operating across EU and US.
| Need | What you use today | VERA.AI adds |
|---|---|---|
| Data labeling | Encord, Labelbox, Scale AI | Connector, no replacement |
| AI governance | Credo AI, Holistic AI, OneTrust | Connector, no replacement |
| Model training | SageMaker, Vertex AI, Azure ML | Metadata import |
| Annex IV documentation | Manual Word/PowerPoint, 6-9 months | Auto-generated in days |
| Audit chain proof | Nothing (or claims only) | SHA-256 hash chain, verifiable |
| PII pseudonymisation | Manual or none | HMAC-SHA256 automatic |
| Multi-framework (EU + US) | Two separate projects | One platform, both packs |
Building Annex IV documentation for a complex AI system takes 6-9 months. Starting in 2026 is normal procurement timing for a 2027 deadline.
Article 5 prohibitions and Article 50 transparency obligations remain enforceable from August 2026 regardless of Omnibus.
Banking AI, insurance AI, hiring AI, public sector AI all require complete Annex IV documentation. No exceptions.
AI in medical devices, automotive safety, industrial machinery, and other regulated products.
Free assessment: submit one AI system, we deliver a complete Annex IV readiness report and a sample evidence pack. No commitment.
Request your free assessment