Services for AI developers
Build AI that is safe, defensible, and ready for regulatory scrutiny.
We loved working with technical teams - and translating priorities between product, compliance, legal, and security. These services are designed specifically for companies developing AI and agentic AI. See a customer testimonial from Intric AB to the right.
Developer‑focused services include:
-
Horizon scanning of upcoming regulatory needs
-
Compliance mapping and documentation: EU AI Act articles 9-15, 13, 27, 50, 53-54, 55, and 73
-
Translation of your responsible AI practices into customer- and market-facing assets
-
Governance of AI agents
-
Recommendations for in-product compliance support
-
Customer-facing training and communication
-
Bias evaluation (see below)
-
Preparation for EU conformity assessments
Best for:
-
Vendors selling into public sector or regulated industries
-
Start-ups and scale-ups
-
Companies wanting to focus on responsible AI as a competitive advantage
Multiple formats are available:
Technical advisory • Pairing with your tech teams, or more client-facing • Project-based
Winning tenders
Public competitions are beginning to require compliance with the EU AI Act, in addition to GDPR.
If you are selling to the public sector, we can help you put together a strong compliance documentation package. Compliant and responsible AI can be your competitive advantage, if you can demonstrate this clearly.
If your system processes personal data and your potential customer is asking for a preliminary DPIA, as many public organizations are, we can review your DPIA and corresponding risk assessment.
This service is closely connected to the next, which is assessing your system's overall compliance.
EU AI Act compliance
We help vendors meet EU AI Act compliance requirements, end-to-end, from UI transparency notices to obligatory conformity declarations.
Transparency requirements for many AI systems are already live, as are AI literacy requirements. High-risk AI systems have until 2027.
We can help you assess your AI system's risk classification, map every single of the hundreds of AI Act requirements against your current system or design, and advise you on which harmonized standards to purchase and use.
Harmonized standards are voluntary, but incredibly powerful tools for you to use. If you follow them, your high-risk AI system will be presumed to be in conformity with the corresponding AI Act requirements. (Ley is an appointed European expert to CEN/CENELEC's Joint Technical Committee 21, which is responsible for developing the nine harmonized standards for high-risk AI systems.)
Many vendors' customers also have their transparency requirements, particularly if customers are public organizations or if your system is being used in a way that affects customers' employees. We are available to develop customer-facing transparency and compliance packages, to increase customer confidence and help them meet their own obligations.
Ethics, risk reviews and impact assessments for developers
Fundamental rights impact assessments (EU AI Act article 27), human oversight (article 14), and claims of exceptions (article 6).
We have a unique AI ethics review methodology. This identifies and involves key stakeholders such as your end users who are historically likely to be negatively impacted, includes them in participatory risk brainstorming, and helps you create AI impact assessments and AI ethics reviews that demonstrate market leadership.
Fairness auditing for developers (prEN 18283 and 18284)
Given high-profile failures of popular AI systems, fairness and non-bias can be your competitive advantage.
One of our unique services: a methodology to help you document whether your AI system is “fair” (or "responsible", or other less commonly quantified values), in collaboration with your developers. These results can give you the metrics to demonstrate safety and responsibility to the market.
These audits are directly informed by my role developing CEN/CENELC's harmonized standards on bias evaluation (prEN 18283 and 18284), which will presume conformity with the EU AI Act Article 10.