A small studio building custom AI systems for a handful of brands each year. Trained on your voice, shaped by your craft, owned entirely by you.
- Maison Veroli
- Pirelli Studio
- Lavazza Lab
- Brunello Cucinelli
- Sant'Ambroeus
- Olivetti Renovatio
- Boggi Milano
- Confindustria
The market is full of AI agencies. This isn't one of them.
We started Primaforma because we noticed something quietly absurd. Brands were being handed AI tools trained on the public internet — tools that had never read a single sentence from the brand itself. A model that has ingested everything has learned nothing about restraint.
A model trained on the public internet doesn't speak the way Cucinelli speaks, or write the way Brunello's editorial team writes, or reason the way a fifty-year-old marque reasons about colour and season. It sounds like everyone. It sounds like no one.
So we built a different kind of studio. One that starts with listening. One that treats AI training as a craft process — iterative, slow, precise — not a configuration step.
What we deliver is yours. The weights, the infrastructure, the fine-tuned instincts. Not a subscription. Not a dependency. A commissioned object, built to your specification, handed over in full.
Manifesto · Method
Fraunces · Inter Tight
A model that writes like the founder.
A sommelier that remembers every vintage.
The voice that travels with the product.
Forty years of craft, made searchable.
Teaching a model to write with restraint.
Veroli's editorial team writes about objects that exist between architecture and fashion. We built a model that learned to do the same — without explaining itself.
Read the case →Reduction in editorial cycle time, measured across two seasons.
Of brand archive ingested, cleaned, and structured.
Now using the tool daily, with full ownership of weights.
Listen.
We spend the first week reading. Catalogues, internal memos, founder interviews, press, competitor silence. We develop a thesis about what the brand refuses to say as much as what it says.
Architect.
We design the data pipeline, define the training strategy, and build the evaluation framework. Nothing is trained until we agree on what "correct" means for this specific voice.
Build.
Iterative fine-tuning with continuous evaluation against the brand brief. We run fitting sessions — structured reviews where the founders and editorial team pressure-test outputs against real use cases.
Deploy & Own.
We hand over the weights, the infrastructure, and the documentation. You own everything. We train your team. We are available for 30 days of post-delivery support.
A brand is not data. It is a refusal.
A list of things it would never say,
worn smooth by years of saying them.