There is a family story Ilya Gorelik has heard many times. He was three years old, running along a beach in Ukraine, shouting the same words again and again: "I am a Jew." His mother tried to quiet him, but the boy had already grasped something that would shadow him for years — he did not look like the people around him, and he would have to learn to see the world with unusual clarity.
That clarity hardened into a discipline.
In the family where Ilya grew up, money was never the highest value. Honesty mattered. Discipline mattered. Hard work mattered. But above all of it sat one thing: critical thinking. A child was expected not only to act but to explain why. Reasoning was not decoration; it was a condition of daily life.
"I always had to explain the reason I did things," he recalls. The habit never left him.
He was an ambitious, restless child who wanted to master everything in front of him — music, mathematics, English, competitions. By the age of five he had decided he would become a mathematician, and he imagined his future with the certainty only children and true believers possess.
Then came the first of the failures that shaped him. His dream was to study mathematics at Moscow State University. Ilya knew that admission was nearly impossible for a Jewish student in the Soviet Union, but he prepared for years anyway — and, like so many other Jewish applicants, he was rejected. It was a heavy blow. He was crushed, he says, "like a bug." Instead of ending his ambition, the rejection hardened it. He earned a Ph.D. in mechanical engineering at twenty-five — partly through luck, he admits, but mostly because he was determined to prove he was somebody.
That mixture of ambition and discipline carried him into technology. He founded Deltix, a quantitative trading platform later acquired by EPAM Systems; served as a CTO at Parametric Technology Corporation; led AI innovation at EPAM; and built enterprise AI platforms through DialX.
But his path was never a clean story of rising success. At one point, after helping take a company public, he was pushed out of the very organization he had built. He lost money. He felt betrayed. And he began to question the kind of loud, visible success he had been chasing. The experience changed him — he stopped needing to be seen and started thinking harder about the principles he actually wanted to build on.
So when generative AI arrived, Ilya recognized something familiar: brilliance tangled up with hype and unpredictability.
The models were impressive. They could answer questions, write explanations, analyze data, and produce confident conclusions on demand. But in enterprise settings, hallucination and the absence of determinism were not acceptable. The reasoning was usually hidden. The logic was often impossible to inspect. And people were already starting to trust answers they had never checked.
For Ilya, the frightening part was not that AI could be wrong. People are wrong all the time. What worried him was quieter: that people would accept the machine's answer without ever questioning it.
"The dangerous thing is not that it makes a wrong decision," he says. "The dangerous thing is that you don't verify it."
That conviction became one of the foundations of DistilX, which he co-founded with Aliaksei Vavilau, an engineering leader who had spent years building mission-critical systems alongside him. The premise is blunt: in high-stakes domains, "good enough" is not good enough. A serious answer has to become a decision the organization can inspect, validate, repeat, and defend.
A bank weighing credit risk, an insurer pricing exposure, a hospital flagging patient risk, a car-retail platform ranking sales leads — none of them can afford "the model recommended it" as an explanation. The logic has to be visible. The assumptions have to be tested. The human expert has to stay in control.
DistilX turns AI reasoning into deterministic, executable, auditable business logic. A domain expert describes a problem; the system lays out a reasoning path; each step becomes visible and tied to the company's data. The expert can challenge it, approve it, and deploy it into production for reuse. At the end, the organization receives a Correctness Validation Report documenting exactly how the decision was reached — and carrying a human stamp of approval.
But when Ilya describes the product in the plainest terms, he doesn't reach for enterprise architecture. He calls it, almost playfully, "a gym for the brain."
It is a system designed to make people ask why they know what they know. It takes the best of AI — breadth, speed, reasoning — and binds it to validation, verification, and deterministic computation. What it refuses to do is let the user's brain go lazy.
That phrase may be the truest way to understand both the company and the man behind it. Ilya is not trying to make AI sound magical. He is trying to make it accountable. A
lifetime has taught him to distrust easy answers, to question authority, to outlast rejection, and to demand reasoning where others settle for confidence.
Now, as companies rush to wire AI into real business workflows, that discipline is starting to look less like skepticism and more like foresight. The first wave of AI dazzled the world with answers. DistilX is betting that the next wave will belong to decisions that can survive scrutiny.
And perhaps that is why his story matters. He is not simply building another AI company. He is building a technology shaped by a lifetime spent asking one stubborn question:
How do we know that this is true?
