We didn't want to ask permission.
It should be as straightforward to run AI in Stockholm as in San Francisco. We refused to accept that it wasn't — so we built it.
The problem
Every company in Europe wanting to build with AI today ends up in the same spot: a handful of American APIs, terms that can change overnight, data sent to a jurisdiction where the GDPR doesn't apply. Open-source models exist, but self-hosting them takes GPU expertise few teams have time for.
We kept seeing the same pattern: good ideas stalling at the prototype stage because the step to production involves either a long-term lock-in to a non-European provider or months of infrastructure work.
The solution
One unified API. OpenAI-compatible so your code moves over without changes. You send your request, we run it in Europe, you get the response. No other choices to make.
All data stays inside the EU/EEA. All accounting follows Swedish rules. A cash invoice is issued automatically on every top-up. The goal is a stable API surface over time, with clear versioning when breaking changes are needed.
What we stand for
- Openness first. We run GLM 5.1 (open-weight, MIT license) on our own hardware — you can download the weights and run them yourself if we slack off.
- Europe. Data stays in the EU/EEA. Nothing is sent across the Atlantic without you knowing.
- Transparent pricing. No hidden fees, no markup on tokens beyond what we state. Receipts follow the Swedish Bookkeeping Act.
- Simplicity. One API, one account, one clear price. If you need more it's because you grew — not because we engineered complexity.