Ship high-risk AI that proves itself
InvGuard detects invoice fraud. rify-AI anchors every decision cryptographically. The Compliance Toolkit auto-generates the EU AI Act paperwork — with tamper-proof evidence nothing was altered after the fact.
- Annex III–ready
- Tamper-evident trail
- Auditor-ready PDF
The compliance-by-design flow
InvGuard
Invoice → fraud decision + explanation
rify-AI
Hash · sign · anchor on-chain + IPFS
Compliance
Model card · bias · audit report
€50B+
B2B invoice fraud / yr in the EU
40→22%
False positives vs. rule-only
100%
Decisions cryptographically verifiable
2026
AI Act enforcement begins
Built for the teams on the hook for AI Act conformity
One platform, three components
Compliance grounded in a real pipeline
Most AI-governance tools are wikis bolted onto a dashboard. This one sits on top of a production model, so the documentation and the proof are real.
InvGuard
A real high-risk fraud engine — not a toy notebook. Explainable by design.
- Versioned rule engine + graph heuristics
- Per-decision explanations & feature weights
- FastAPI scoring in under 5s
rify-AI
Every decision is committed with a canonical hash, signed, and anchored.
- Reproducible canonical-JSON commitments
- On-chain record + IPFS context
- Independent tamper verification
Compliance Toolkit
Auto-generated AI Act artifacts, grounded in your real pipeline metadata.
- Templated model cards (no hallucinated claims)
- Proxy-attribute bias reporting
- Auditor-ready combined report
How it works
Invoice in, defensible evidence out
01
Submit an invoice
InvGuard scores it and returns a decision, risk level, and a full explanation.
02
Anchor the decision
rify-AI hashes, signs, and anchors it — the record becomes tamper-evident.
03
Generate the report
The toolkit assembles the model card, bias report, and audit PDF from live data.
Social proof
What practitioners want to hear
Illustrative of the buyer personas this platform is built for.
We stopped re-typing model details into a governance wiki. The report is generated from the pipeline we actually run.
Head of Compliance
Mid-market B2B payments
Being able to hand an auditor a decision they can verify themselves — not just trust our database — changed the conversation.
DPO
EU fintech
Fewer false positives than our rule engine, with an audit trail attached to every call. That combination is rare.
Head of Fraud & Risk
B2B marketplace
About
Compliance shouldn't be a wiki bolted onto a dashboard
The EU AI Act puts real obligations on anyone running high-risk AI — fraud and credit decisioning included. Today that means expensive manual documentation, usually outsourced to consultants who don't understand the model.
We took the opposite approach: build a genuine high-risk system first, then wire it directly into the documentation and monitoring engine, and anchor every decision so the audit trail is cryptographically tamper-proof. Real system, real paperwork, real proof.
Compliance-by-design
Docs derived from the live pipeline, not re-entered.
Tamper-evident
Canonical hash + signature per decision.
Honest scope
Heuristics today, trained GNN on the roadmap.
Configurable
Questionnaire content updates without a rebuild.
Contact us
See it on your own data
Tell us about your high-risk model and we'll walk you through a scripted end-to-end demo.
- hello@compliant.eu
- Based in
- European Union · GDPR-aligned
- For auditors
- Request read-only verification access to a sample decision set.
Ready to make your AI defensible?
Run the full loop — invoice in, decision anchored, report generated — in a single walkthrough.