Only what's needed
The system only gets the data it needs for that one task.
AI Agency
Security and privacy. Agreed up front, not after the fact.
Before we build anything, we agree together on what data the system uses, where it runs, who has access and what a person needs to approve.
How we work
Six commitments on every project.
No training on your data
For the AI we use Anthropic's business API (Claude). Under Anthropic's business terms, your data isn't used to train their models.
You know where it runs
Agreed per project. If it has to stay within the EU, we run the model in a European region, for example through Amazon Bedrock in Frankfurt.
Access and audit log
Who has access is agreed up front. The system keeps track of what it did and why.
A person approves
Nothing goes to a customer until someone on your team approves it.
Data processing agreement
We sign one with you, listing the services we use.
EU AI Act
What the EU AI Act asks of you.
The European AI Act also applies to small businesses that use AI. These are the points that relate to our work.
- People know it's AIAnyone talking to a chat assistant or voice agent has to know it (Article 50, starting Aug 2, 2026). Our assistants say so right at the start.
- Your team knows how it worksAnyone working with AI needs to know what it can and can't do (Article 4). You get a short walkthrough and a written guide.
- No decisions about peopleWe don't build systems that select or assess people, such as for job applications or loans.
FAQ
Questions we get a lot.
Do you use my data to train AI?
No. And neither does the AI service we use: that's in their business terms and in our data processing agreement.
Where is my data stored?
We set that out per project in the proposal, including the services we use and which country they're in.
Can I see what the system did?
Yes. There's a log of what happened, when, and why something was sent to you.
What happens if I stop?
We switch it off, hand over what's yours and delete your data on our end.
Related services