You know, I’ve been thinking for a long time whether I should tell this story. But maybe it will be helpful for someone to learn from others’ mistakes. Especially if you suddenly consider buying a small business or launching your own startup.

It all started very positively. I met a guy named Thomas at Hong Kong Airport — a typical story: flight delay, helping with a taxi and some cash. Then he offered us a project: a system for predicting stock prices based on analysis of exchange documents. Sounds ambitious? Yes, it was.

Thomas had just graduated from a prestigious university, eager to prove himself, and was basically a typical young businessman with his dad’s money. By the way, his father was a well-known Hong Kong businessman, Lu Cheng. The contract was for half a million dollars. Seemed like a jackpot.

But here’s the catch. The pilot phase went smoothly — we worked with clean data from the Singapore Exchange, everything was structured. Then it turned out we needed to process documents from 28 exchanges, with different formats, crooked scans, stamps over text, handwritten entries. It’s like thinking buying a small business is easy, and then discovering you need to redo all the wiring and plumbing.

We honestly told them about the problems. Proposed to rearrange the stages: first handle the simple cases, then the complex ones. And that’s when the miracles began.

Katrin appeared — Thomas’s employee, or so it seemed to us. She was the link between us and the client. But then it turned out she was working not just for Thomas, but for the entire Cheng family. And the family had their own plans.

Payments started to be delayed. Then complaints. Then Thomas suddenly “lost it” — literally in the middle of the project, they decided to move everything to a cloud service. Who convinced him? Two Brits, Bob and Mike, representing a well-known cloud provider. Coincidence? Hardly.

We started paying out of our own pocket. Burn rate — watching your money go to pay the team while no income comes in. It’s a hellish experience.

Then I flew to Hong Kong for a meeting. The office was in boxes — either moving or shutting down, it was unclear. Thomas wasn’t there. Instead, I met with Katrin. She laid out an audit from a major IT company, accusing us of using wrong technologies, wrong methodologies, wrong everything. Then she pulled out a lawsuit. And at the end — tore our contract in half.

“The Cheng family only pays for 100 percent, not 93 or 95,” she said. And text recognition is always probabilistic, never 100 percent. It’s like demanding a guarantee that you buy a small business and it will immediately be profitable.

Then it turned out Thomas “lost his mind” from the luxurious lifestyle. He was suspended and sent for treatment. Accounts were frozen. Katrin also got it — demoted to security. They hired an Indian team, but they disappeared. The project was transferred to a new owner who simply bought the exchange data feeds and buried all our work on recognition.

The losses were comparable to the value of a couple of apartments in Moscow. Had to lay off people. Spent three years paying off debts.

And you know what’s funny? Now everyone talks about AI and document recognition. But no one has shown 100-percent recognition. Because it doesn’t exist. It’s physically impossible.

The takeaway: if you’re thinking about buying a small business or taking on a big project — check who’s really making the decisions. Don’t rely on the pretty promises of young entrepreneurs. And remember, there are no perfect technical solutions. Reality is always more complicated than what you see in PowerPoint.