Brembo has been casting and machining brake components since 1961. F1 calipers, the discs on the supercars not many can afford. Heavy industry, molten metal, the opposite of a software startup. Then in October 2024 it started selling software: ALCHEMIX, an AI tool that generates new material recipes from a set of performance targets and raw-material constraints. Formulating a new compound used to take days, sometimes weeks of trial and error. ALCHEMIX does it in minutes.
Brembo is known for their excellent brakes, ready to race.
The product they never meant to sell
They never built ALCHEMIX to sell. They built it for themselves, to speed up their own R&D on brake pads, and it worked so well that customers and partners started asking to use it. So Brembo spun it into a business, Brembo Solutions, with its own P&L, and put ALCHEMIX on the Microsoft Marketplace as a cloud service. Its first paying customer was not a car company. It was a premium food and beverage maker. It now sells into cosmetics, chemicals and textiles too.
A company that makes brakes now earns money from a food company. The product is not a brake but six decades of knowing how to get a material recipe exactly right, packaged as software.
Screenshot of ALCHEMIX from the Microsoft app store.
Cost line, or product line?
Most owners treat AI as a cost line: a way to shave time, trim waste, do the same work with fewer people. Fair enough, it does that. But for a company that makes something, an AI tool or agent can be a product line too. What you actually own is more than the physical product. It is the accumulated know-how of making it well, the judgement your best people carry in their heads, the process nobody else has. For all of industrial history that know-how was almost impossible to sell. You could sell the brake, barely the process of making it. AI is the technology that lets you package the knowing and put a price tag on it.
The know-how also travels further than the product does. Brembo's team found that once you get down to the level of data and process, a brake maker and a biscuit maker speak the same language. The formulation problem is the same shape. So the tool built for friction materials works, with adjustment, for chemicals and cosmetics and food. The moat you spent sixty years digging in one industry becomes a product you can sell into ten.
Price range for the Brembo’s app might make you jealous.
The jealousy audit
How do you define what can be 'servitised'? Which of the things you already do better than anyone would someone outside your industry pay for?
Step 1: list the internal tools. Every automation, model or clever spreadsheet teams built to make their own work faster. Not the ones they bought. The ones they made, because those encode know-how nobody else has.
Step 2: mark the ones that would make a stranger jealous. For each, ask whether a competitor, a supplier, or a business in an entirely different industry would pay for that output.
Step 3: give it a try and sell it. If there's a market, you'll see the revenue.
Brembo found a way. Starting a SaaS business is becoming democratised?