How we work
From strong open models to AI built for your work.
Every product and service we offer follows the same path — from proven base models to our own models, shaped by our team for the people and businesses who use them.
Our approach, step by step
Start from strong foundations
We start from leading open-weight base models, released by established model makers.
Make them our own
We will post-train these base models into our own models, shaped for real work.
Serve them as our own
Our post-trained models will be deployed and served as our own model service, ready to answer requests.
Power our products
Our own products, CodeFlower and Collacom, will run more and more on these models.
Build for businesses
We design custom AI applications and agents for businesses, connected to their systems and built to run on our models.
Open to more industries
Next, we will offer industry models — and models customised for a single company — through our API.
Our own GPU facility near Milan
Our own GPU facility is planned in the Milan area. Our GPU servers will be used for our own post-training and for serving our own models, in Italy.