Lavazza Group · NIMS
About 100,000 leads. Nobody learned a new tool.
~100,000 leads
Read the caseLoading model
Your competitors have ChatGPT too. We build you an AI they can't buy.
339 clients in 15 years, 135,000 leads and 708 million impressions in the campaigns we run. Now the AI: one company at a time, reading your stock, your price lists, your CRM, on your own machines.
| PROCESS | WHERE | RATE |
|---|---|---|
| scintilla | MacBook Pro | 166 tok/s |
| imagineer | MacBook Pro | 1.4 s / image |
| eugeny | MacBook Pro | 27B in 7.6 GB |
| tomasi.feed | 5 showrooms | 850 pages |
| nims.leads | Meta · Google · TikTok | ~100,000 leads |
| pam.app | iOS · Android | 713,836 installs |
Pam Perte Plus
The supermarket in your pocket.
713,836 installs
Read the case
Viessmann
The job goes to the nearest installer who knows the product.
52% to inspection
Read the case
FIAMM Energy Technology
We make batteries worth following.
29,744 LinkedIn clicks
Read the case
AIA · Wudy · Pam · Damigella Blu
Then somebody has to shoot it.
233 productions
Read the case
Pam Panorama
This week's flyer, tonight's dinner.
12 recipes from 1,000
Read the case
Tomasi Auto
The stock writes the ads.
850 pages
Read the case€13.3MTwo months
Cars sold to customers acquired through digital channels.
Automotive
From digital enquiries to car sales.
€13.3M in two months
Read the case
Qualified leads, delivered to your CRM.
See howMore revenue.
Not words.
Not words.
We turn ad budgets into real customers.
- coffee · direct sales
- supermarkets
- heating
- food
- entertainment
- car dealers
- banking
- banking
- consumer goods
- consumer goods
| clinics | dental and aesthetic practices: the requests that arrive in the evening | 135,000 leads |
|---|---|---|
| professional-firms | law and accounting firms: every request to the right person | 9,400 appointments |
| real-estate | estate agencies: valuation requests with the property's data filled in | 34,350 requests |
| italian-versions | global brands: the Italian version of your campaign | 233 productions |
| stockists | fashion: the shops that sell brands like yours, and who buys | 2.29M ads read |
| lead-generation | qualified leads: every item in your stock gets its own page and its own ads | 135,000 leads |
|---|---|---|
| creator-campaigns | creators cast, shot on our set and run as paid media | 233 productions |
| custom-ai | one process, in production, on your data and running where you decide | 713,836 installs |
AI made to measure for you.
Efficient enough to run on your laptops: no server racks, no data-centre graphics cards. Our models work on the Macs you already have, even with the Wi‑Fi off: on a train, at a client's, in a plant with no signal.
Describe the software. Get it live.
fornace.net turns fifteen years of making sites and apps into a builder anyone can run: a playable voxel world, a sprint-training app, a property research tool and Bonazzi's wine shop were all built there.
Build something on fornace.net ↗
Models and engines built in Valpolicella.
Trained, compressed and run on our own C, Rust and Metal engines, down to a laptop chip.
| scintilla | language model with its own engine, runs on any Mac with an Apple chip | 166 tokens/s on a MacBook Pro |
|---|---|---|
| eugeny | a big model, compressed to fit on a laptop | 27B parameters in 7.6 GB |
| imagineer | images and video on a Mac, offline | one image in 1.4 s |
| torchio | compresses models to 2 bits and keeps what they know | 2.2× more accurate than plain 2-bit rounding |
| forgia | where our models get their data, their training and their exams | every step on record |
| gateway | one door to every AI model, with budgets per person and automatic failover | open source |
| matchbox | identity, permissions and secrets, so nobody carries a password | 21 connectors |
| fornace.net | describe the software, get it live | live |
MacBook Pro M5 Max, 48 GB. Imagineer: 512 px image, 4 steps, warm. Scintilla: q8, short context.
Everyone gets AI. Nobody gets the keys.
Your colleagues sign in with the company account and each finds the AI their role allows: sales the CRM, accounts the ledger, the new hire an onboarding guide to ask anything. Every model behind one door, every action logged, no password in anyone's hands.
09:02 giulia sales crm.search ok09:03 marco new hire onboarding.ask ok09:05 agent:ads marketing meta.upload approved by luca09:06 marco new hire ledger.read denied
20 people and a 50 m² set in Valpolicella.
Next to the office in San Pietro in Cariano sits our 50 m² kitchen set with cinema cameras, lights and drones, where we shoot Viessmann's 4K spots and NIMS's product films.
Fornace was founded in 2011 by Francesco Frapporti and Luca Cisorio. It began with native iPhone apps for marketing. Today it's 20 people plus a circle of collaborators, in Valpolicella.
Start with a call and a measured pilot.
- Call. You tell us where time or money leaks. We start where AI fits your process.
- Pilot. One process, one number to move, agreed before we write a line.
- Yours. Model, pipeline and permissions, running where you choose, your colleagues' laptops included.
Book a call Send this to your IT
“We already pay for ChatGPT Enterprise.”
Keep it. It has never seen your stock, your price lists or your CRM. We build the part that has.
“Isn't this ChatGPT with your logo on it?”
No. Scintilla, Eugeny and Imagineer are ours: we train them, compress them and wrote the engines they run on. When we use someone else's model, it goes through the same door as ours.
“Will you build the same thing for our competitor?”
Not with your data. The model trained on your work, and the code around it, belong to you. Your training data and every answer the model gives stay in a chain of custody we sign with you. It's in the contract.
“Does our data end up in the US?”
Only if you send it there. Our models run at your premises, in approved European data centres, or on one laptop with the Wi‑Fi off.
“Do we need servers with graphics cards?”
No. To start, a Mac from a shop is enough. If you'd rather keep it in a rack, we host at your premises or in approved European data centres, with the chain of custody in writing.
“We're not a multinational.”
Neither is Tomasi. Five showrooms, 850 pages, and the stock keeps them current.
“How is quality guaranteed?”
Code checks what code can check, and a person signs off what goes out. At NIMS, AI never produces a number: the code counts the leads.
“What about the AI Act?”
Before we build, we write down which obligations fall on you and which on us. Generated content is labelled, every action is logged, permissions are set per person. When your lawyer arrives, the answers are already written.
“Can we leave?”
Yes, and you take everything with you: models, data, code. Whoever comes next starts from there, not from zero.
“Our IT will have questions.”
Here is the technical sheet for them: where the models run, how Matchbox holds secrets and how exit works.
“How much?”
Scope and fee are fixed in writing after the first call.
Write to us. Francesco or Luca answers.
Usually Francesco or Luca.