Stockholm
Sweden
Our home lab. Where the models get designed, argued over and rebuilt — and where Datapect, Smart Stock Data, Visuasic and TaxGuru went from a whiteboard to production.
- Applied research
- Model development
- Product engineering
Our Labs
Not offices.
Laboratories.
We call them labs because that is how they work. Every product we ship started as an experiment that someone was allowed to run — and most of the experiments that failed taught us more than the ones that worked. Two locations, one practice: read the frontier, build the thing, put it in front of real data, and keep only what holds.
Abu Dhabi
United Arab Emirates
Our lab closest to the people using what we build. Research gets pressure-tested against real operations here, and comes back sharper for it.
- Client delivery
- Data platforms
- Regional partnerships
How a lab works
Research that has to survive contact with production
01
Read the frontier
We track what is actually landing in AI and data — not what is trending. If a technique cannot beat the simple baseline on your data, it does not ship.
02
Prototype in days
An idea earns its next week by working on a real dataset. Small, fast, disposable experiments beat long specifications.
03
Harden it
What survives gets pipelines, monitoring, evaluation and the unglamorous engineering that turns a notebook into a product people trust.
04
Keep measuring
Models drift and data changes. Every system we run keeps reporting on itself, so we find out before you do.
Got a problem worth
experimenting on?
Bring it to whichever lab is closest. We will tell you honestly whether AI is the right answer — and what we would try first.
Talk to a lab