Designed and hosted in Europe
AI compliance isn't declared. It's demonstrated.
AURA traces the classification of your AI systems, checks AI Act requirements, and measures bias as well as carbon footprint. Every finding is tied to timestamped evidence secured by a cryptographic fingerprint, verifiable by an auditor. All without exposing your data.
Full compliance for your AI systems
AURA examines your models along three critical dimensions, for AI that is ethical, compliant and sustainable.
Legal compliance
Our solution analyzes your models against the main European regulations: AI Act, GDPR, NIS2, etc.
It gives answers grounded in the official texts, with full source traceability, backed by a sovereign language model.
Answers that cite their sourcesAlgorithmic bias
We evaluate generative models on 6 criteria: gender, age, disability, ethnic origin, religion, and sexual orientation.
We also audit the embedding models used in your solutions: search engines, CV matching, internal RAG, etc.
In local mode, your models and your data never leave your infrastructure.
6 generative categories, 3 embedding categoriesCarbon footprint
We precisely measure the real energy consumption of your models, comparing their consumption at rest and under load, to assess their environmental impact.
Measured at every callThree audit levels to suit you
A step-by-step, modular approach for every kind of company
Quick assessment
A 22-question questionnaire to situate your AI Act risk level. Immediate result, with a colour-coded summary.
Free · 10 minutesDocumented analysis
Chatbot with legal retrieval for in-depth analysis. Follow-up questions, and answers anchored in the regulatory texts, fully traceable.
On quotationFull audit
Adversarial testing across the three audit axes. Detailed report with prioritised recommendations. On-site intervention or API access, as you prefer.
Premium · On quotationSolutions matched to what you need
Choose the audit level that matches what is at stake for you
Free
- A 22-question questionnaire
- AI Act risk score
- Immediate result
- Penalty estimate
- Email support
Professional
- Legal AI chatbot
- Retrieval over legal texts
- In-depth analysis
- Full traceability
- Detailed PDF report
- Priority support
Enterprise
- Full three-axis audit
- Adversarial tests
- HELM bias analysis
- CodeCarbon CO₂ measurement
- On-site intervention
- Dedicated support
- API integration
The European reference for AI auditing
A sovereign, precise and ethical route to full compliance
European sovereignty
European LLMs (Mistral, EuroLLM), EU hosting, GDPR compliance by construction. Your data stays in Europe.
Provider of your choice
Mistral by default, or the provider of your choice — down to your own self-hosted model, if nothing may leave your infrastructure.
Leading-edge technology
Recognised methods, not in-house estimates: the BBQ metrics from the HELM benchmark for bias, the CodeCarbon methodology for carbon. A measurement you can verify, not a promise.
Eco-Designed Architecture
No AI models hosted in-house, no oversized infrastructure: AURA relies on the providers of your choice, called on demand. A platform frugal by design.
Multilingual
French, English and Dutch. Legal terminology follows the official wording of the Regulation in each language.
ISO 42001 certification
ISO 42001 certification for AI management systems is under way.
A solution for everyone working with AI
Compliance officers
Avoid AI Act penalties. Obtain complete, traceable documentation of your AI systems' compliance, with detailed reports ready for the authorities.
IT directors and CTOs
Secure your AI deployments. Rigorous adversarial testing, environmental-impact measurement, and technical validation of your models before production.
Lawyers and legal counsel
Thorough legal documentation. Precise references to the articles of law, fully traceable analyses, and reports ready for your client files.
Start your AI Act compliance audit
Join the European companies choosing AI that is compliant, ethical and sustainable. Situate your risk level free of charge, in 10 minutes.