The EU AI Act is already in force
Regulation 2024/1689 applies in phases since 2024. Prohibited practices and AI literacy duties already apply, and further obligations follow on a fixed calendar.
ISO 42001 · Paris
In Paris, we build your AI management system to ISO/IEC 42001:2023, from AI inventory to certification audit. You get AI governance that auditors, customers and the EU AI Act can all rely on.

Local context
Many Paris software publishers now ship AI features inside their products. That makes them AI providers under the AI Act, with duties toward their customers. Enterprise buyers in La Défense and across Europe now ask for an AI governance answer next to ISO 27001 or SOC 2. ISO 42001 gives that answer a form an auditor can certify.
In France, the CNIL has published recommendations on developing AI systems that process personal data. We fold them into your impact assessments and data quality controls. Banks, insurers and asset managers supervised by the ACPR or AMF already run model governance processes. We link your AI register to those processes instead of building a parallel one.
Our team is based in Brussels, not in Paris. We run most of the work remotely, from inventory interviews to document reviews. We then plan on-site days for workshops, the internal audit and the management review. Paris is about 1 h 25 from Brussels by high-speed train, so each of those days fits in a single trip.
Paris and La Défense host financial institutions, insurers, software publishers, start-ups and public bodies. French supervisors are active, and many Paris companies sell to European customers who require ISO 27001, SOC 2 or DORA evidence.
Paris is about 1 h 25 from Brussels by high-speed train. We combine remote work with on-site days for workshops, interviews and audits.
What is at stake
Regulation 2024/1689 applies in phases since 2024. Prohibited practices and AI literacy duties already apply, and further obligations follow on a fixed calendar.
Procurement teams now send AI questionnaires next to security ones. Without a documented system, each answer is improvised and deals slow down.
Bias, opacity, model drift and misuse are not covered by a classic security risk register. They need their own assessment and owners.
Key sectors in Paris
Banks and insurers use AI for credit, fraud and pricing, and ISO 42001 gives supervisors a clear view of how those models are governed.
Software vendors embedding AI face customer questionnaires, and certification answers them once instead of deal by deal.
Public bodies using AI for citizen services must show fairness, transparency and human oversight, which ISO 42001 turns into documented controls.
Firms using generative AI on client work need clear rules on confidentiality, review and disclosure, set out in an AI policy.
What we do
We do not start everything at once. The scoping call shows which workstream matters most, and we start there.
We list every AI system you build, buy or embed, including third-party APIs. We confirm your role for each: provider, deployer or user.
We extend your risk method to AI-specific harms. We run the AI system impact assessment required by the standard for each in-scope system.
We write the AI policy, roles and lifecycle procedures from design to retirement. Controls cover data quality, transparency, human oversight and suppliers.
We run the internal audit and the management review, then prepare your teams for the certification body. If you hold ISO 27001, we extend that system rather than duplicate it.
A 30-minute call is enough to scope your AI management system and its next steps.
Regulators
French national cybersecurity agency, competent authority for NIS 2 in France.
French data protection authority: GDPR supervision, audits and sanctions.
Financial supervisors for banks, insurers and asset managers, including DORA.
GDPR, DORA and the EU AI Act apply directly in France, as in Belgium.
The method
Each step ends with a document you keep, even if the engagement stops.
A free 30-minute call to understand your AI use, your customers' demands and your timeline.
Deliverable : Written scope and fixed price per lotWe interview product, data and IT teams and list every AI system in scope.
Deliverable : AI system register with roles and ownersWe compare current practice with clauses 4 to 10 and Annex A of ISO/IEC 42001.
Deliverable : Gap analysis report with prioritised actionsWe assess each system for risks to the organisation, to individuals and to society.
Deliverable : AI risk register and impact assessmentsWe draft policies, procedures and the Statement of Applicability, then help teams apply them.
Deliverable : AI management system documentation setWe audit the system, run the management review and support you during the external audit.
Deliverable : Internal audit report and certification readiness fileScope and budget
Questions
The model provider has its own AI Act duties for general-purpose models. You remain responsible for how you integrate, test and present the AI feature to your customers. We document that split in your register and in supplier agreements.
Yes. Paris teams often mix French and English, so we deliver documents in either language or both. Interviews and workshops follow the language each team uses day to day.
Yes. We work on-site or remotely, in French, Dutch or English. Workshops with product and data teams often work best in person.
No. Certification is issued by an accredited certification body, which must stay independent from the consultant. We prepare you and support you during the audit.
No. ISO/IEC 42001 is a voluntary standard. It is often requested by customers and gives a structured way to meet EU AI Act duties. Certification shows a third party has checked your system.
Last step
Tell us how you use AI today. You receive a written scope and a fixed price per lot after our call.