CUJO, CAST AI, and now CUBE AI: the same pattern. What does Lithuania give you that the Bay Area can't?

Density of serious engineers who actually ship. In the Bay Area you compete with a thousand companies for the same people, and half of them are optimizing their next job hop. In Lithuania you can build a team that stays together for years, which is where compounding really happens in deep tech. The talent pool is small but the ratio of builders to talkers is unusually high. Add a time zone that overlaps with both US mornings and Asia, and a culture that doesn't panic when the problem is hard. The US gives us customers and capital. Lithuania gives us the product.

What have you learned about finding product-market fit? Looks like CUBE AI had a successful pivot.

Three things. First, in B2B you are not looking for people who like your product, you are looking for people whose job depends on solving the problem. If a buyer cannot tell you what happens to them if the problem stays unsolved, there is no market yet.

Second, pivot on evidence, not mood. We started in crypto security, and the technology worked, but the buyer was too small and too fragmented. Banks, on the other hand, lose real money to scams every day and have a regulator watching. Same core skill, completely different urgency.

Third, kill your attachment to the old story fast. The hardest part of a pivot is not the engineering, it is the founder admitting the first thesis was wrong in front of the team and investors. Do it once, do it clearly, and move.

Criminals got AI the same day we all did. Who's winning, attacker or defender? What is your vision with CUBE AI?

Right now the attacker is winning, and it is not close. Scams have become the most profitable cybercrime in history, and AI just removed the last bottleneck, which was human labor. One operator can now run thousands of convincing conversations at once, in any language, around the clock. Defenders are still mostly reacting after the money is gone.

Our view is that you cannot win this by building a slightly better filter on the victim's side. You have to change the economics for the criminal. Every scam ends the same way, with a request to move money somewhere, and that somewhere is the weak point. Cube produces verified intelligence on where scammers collect their money and gets it to banks and payment providers before the victim's transfer clears. The vision is simple: make it so that scamming stops paying, because the payment rails already know where the money is headed.

‍