Ideas that already
outlived the patent.

What is Patent Bank, in one sentence?

JASON VICENTE MEJIA (Berklab) ANSWERS
A bank of ideas: patents that have already fallen into the public domain, reorganized by field and ready to become a product, a service, or research again.

Why do public-domain patents still matter?

JASON VICENTE MEJIA (Berklab) ANSWERS
Because most of them were never built all the way through. The protection expired long before the idea did — a patent from 1948 on solar water heating is just as sound today as it was then, but the company that filed it may not exist anymore, and no one owns the right to stop you from trying it again. What's changed is the cost of trying. Digitized archives make these documents searchable in minutes instead of months, and AI can read a technical description written in patent language and translate it into something a small workshop or a single founder can actually build. The idea was always free; what used to be expensive was finding it and understanding it well enough to trust it.

Who benefits first?

JASON VICENTE MEJIA (Berklab) ANSWERS
The person with the problem, not the person with the capital. A local entrepreneur can open a 1930s patent on drip irrigation and adapt it in an afternoon — work that would otherwise take months of engineering time they don't have. That matters most in places where the newest version of a technology is locked behind a license fee that makes no sense for the scale of the problem. A public-domain patent doesn't care how big your company is. It reads the same whether you're a garage inventor in Lagos, a university lab in Hanoi, or a two-person startup anywhere else — and increasingly, an AI assistant can walk any of them through the adaptation, which levels a gap that used to require an in-house engineer.

And when two patents from different fields cross paths?

JASON VICENTE MEJIA (Berklab) ANSWERS
That's where the site gets interesting. Fusing an energy mechanism with a neuroscience sensor, for instance, usually opens an application neither original patent predicted on its own — a piezoelectric floor tile combined with a biofeedback sensor becomes something closer to an interactive rehabilitation surface than either inventor imagined. Cross-field fusion is also where AI is most useful, because no single person tends to be fluent in both patent archives at once; the model can hold the mechanics of a 1960s ecology patent and a 1980s art patent in mind at the same time and suggest where they actually overlap. We treat every fusion as a hypothesis, not a finished product — the portfolio section documents the reasoning so someone else can pressure-test it, extend it further, or take it in a direction we hadn't considered.