AlphaFold2 solves protein folding

In November 2020, during the CASP14 competition, an artificial intelligence system called AlphaFold2, developed by DeepMind, did something the scientific community had been chasing for nearly fifty years: predicting with enormous precision how a protein folds from its sequence. It may sound distant if your company has nothing to do with labs, but the heart of the matter is deeply relevant to any business.

For decades, that “folding” was considered one of science’s great unsolvable problems. Not because talent was lacking, but because the number of possible combinations was so enormous that no traditional method could keep up. What changed wasn’t the biology: it was the way of attacking the problem. The AI learned patterns from existing data and, with them, solved in hours what used to take years. That same logic is what can be applied today, on a more modest but very real scale, inside a small business.

From an elite lab to your daily operation

It’s easy to think this kind of advance belongs to another world: big budgets, doctorates, and supercomputers. But the principle that made AlphaFold2 possible is surprisingly reusable. It’s about taking a problem that seems chaotic, feeding it the right data, and letting a system find patterns a person couldn’t spot at that speed.

In your company there are “small” problems that feel unsolvable for the same reason: too many variables, too much volume, too little time. Predicting which customer is about to leave, estimating next season’s demand, catching invoices with errors before paying them, or classifying hundreds of emails by priority. It’s not protein folding, but it’s the same family of challenges: hidden patterns in your own data.

What you really need (and what you don’t)

The great news is that you don’t need to imitate DeepMind to get value. What made AlphaFold2 work has accessible equivalents in the day-to-day of a business:

• Data you already have. Sales, support tickets, inventory, customer history. The raw material is usually scattered, but it exists.
• A well-bounded problem. Not “use AI,” but “reduce the time it takes us to quote” or “anticipate which product is going to run out.”
• Custom software. A solution designed for your processes, that keeps your data, your history, and your way of working, instead of forcing you to change them.
• A clear success metric. Hours saved, errors avoided, sales recovered. If it can’t be measured, it can’t be improved.

The lesson of AlphaFold2 isn’t that AI is magic, but that “impossible” problems were often just waiting for the right tools.

The real leap: decisions with reliable information

The most valuable thing about an advance like this isn’t the technology itself, but what it enables: making decisions with reliable information, faster and with less uncertainty. Before, a researcher could spend years without knowing whether they were on the right track. Afterward, they get a solid answer in a short time and devote their energy to what really matters.

In a small business the effect is just as concrete. When AI takes on the repetitive and finds patterns, your team stops putting out fires and starts deciding with a clear head: who to call first, how much to buy, where the money is leaking. Technology doesn’t replace human judgment; it gives it better inputs.

Apply it to your own challenges

You don’t need a giant project to ride this wave. The healthiest way to begin is with a small scope and a goal you can measure:

• Choose a single real pain point. That process that takes everyone’s hours or causes costly errors. One, not ten.
• Organize your data first. Even if it’s in spreadsheets, having it clean and in one place is already half the road traveled.
• Start with a small pilot. A low-risk test, with a clear goal and a short deadline, before scaling to the whole operation.
• Define how you’ll measure the result. Compare the “before” and the “after” with a number: time, money, or errors.
• Lean on someone who translates the technology to your business. The value isn’t in AI on its own, but in integrating it into your processes without losing what already works.

AlphaFold2 showed that a fifty-year problem could be solved when someone dared to attack it differently. Your company doesn’t have to solve biology, but it can apply that same attitude to its own challenges: choose a concrete pain point, measure the result, and build the solution to fit your operation. That’s exactly the conversation we love to have at Normandia Web with small businesses: finding that first hidden pattern in your data and turning it into a faster decision, without you having to give up your way of working.