Security and privacy
Your code. Your rules. Your approval.
Your source code is often the most valuable thing you have. This is how we handle it.
Six security measures
Minimal permissions
Agents only get access to what they need for a task. You can revoke any access yourself at any time.
Isolated working environment
Agents work in a separate environment, apart from your production systems. They cannot put anything live directly.
Human approval
Every change comes back as a proposal. You or your team decide what happens.
Processing in the EU
We choose models and infrastructure with processing in the EU wherever possible. If that is not possible for a particular tool, we tell you in advance.
Not used for training
We only use business services where your code is not used to train models, and we check this in writing.
Data processing agreement
If your code processes personal data, we sign a data processing agreement before we start.
In practice
What we never do.
- Send secret keys, passwords or customer data to an AI model.
- Put changes straight into production.
- Keep your code after the collaboration ends.
- Use tools without you knowing which ones and what for.
Why so strict? In a large test, 44% of programming tasks produced AI code with a risky security flaw. [21] And 45.7% of developers distrust the accuracy of AI output. [20] Human review is no luxury.
At the end of a collaboration, we revoke all access and delete copies on our side. You receive confirmation of this.
Want to know more? Read the article Keeping your code in Europe or see our privacy statement.
What work keeps getting postponed in your business?
Tell us in a free one-hour online conversation. You get a written scan with our honest estimate, even if the answer is that there is little to gain.