Build with AI – the right way.
Many companies already let their teams work with Claude Code, Cursor or Copilot – without clear rules. We help you set up processes where AI ships fast while every line is still reviewed independently.
Where most teams start
- AI tools are in use, but nobody has set rules.
- Code reviews can no longer keep up with AI speed.
- It is unclear which data may go into AI tools.
- The Cyber Resilience Act requires evidence that is missing today.
What we build together with you
Assessment
We look at how your team works with AI today: tools, workflows, risks.
AI policy
Clear rules: which tools, which data, who approves. Short, clear, lived.
Review gates
Automated and human checks in pull requests and before every release – four-eyes principle for AI.
Training & prompts
Workshops for developers and product teams: prompt safely, verify output, spot typical AI mistakes.
CRA evidence
Logs, software bill of materials (SBOM) and review reports – filed cleanly for customers and authorities.
Ongoing support
If you like, we keep reviewing as the neutral authority afterwards – so the processes stay alive.
How a consulting project runs
- 011–2 weeks
Analysis
Interviews, code samples, risk picture.
- 022–6 weeks
Implementation
Policy, review gates, training – right inside your tools.
- 03ongoing
Handover
Your team works on its own, we review neutrally.
Let's talk about your path to AI development.
No obligation, personal. We'll get back to you within one business day.
Not legal advice. For legal questions we can involve specialised law firms on request.