He shut down a startup with real users, applied to Y Combinator six times before it stuck, and turned a year of hesitation into two weeks of momentum. This is how Trace got built.
There is a specific kind of dread that comes from running a company that looks healthy. Tim Cherkasov knew it well. His crypto startup, Clustr, had close to 9,000 users and sessions that stretched near ten minutes. There was even revenue. On paper, it was working. In his gut, it wasn't. For a full year he sat on the feeling that the thing needed to change, and for a full year he did nothing decisive about it.
That gap - between knowing and acting - is the hinge his whole story turns on. Cherkasov is now the co-founder and CEO of Trace, a Y Combinator company building an orchestration layer that decides which work goes to a person and which work goes to an AI agent. To understand why he is qualified to build a manager for software, it helps to know that he spent years learning, painfully, how to manage himself.
Cherkasov started making companies at 17. The industries jumped around - retail tech, then travel, then web3 - which is less a sign of indecision than of a person who treats each attempt as a lesson rather than a lifelong vow. Somewhere in there he picked up an MSc in Business Analytics from University College London and put in real time as a data scientist across hedge funds and scaleups, including the digital-asset firm Nickel.
The most instructive job came at Copper, a digital asset custodian, where he worked as a product manager and delivered an institutional offering for State Street. If you want a founder who understands how large, careful organizations actually adopt new tools, this is the resume line that matters. Enterprises don't buy magic. They buy things that fit into the way they already work. He would need that instinct later.
Clustr was built for people who wanted to invest in crypto without doing hours of homework. It attracted an audience. What it never got was stickiness. Users would read the insights, nod along, and then go execute their trades somewhere else. The value was real but it leaked straight out the bottom of the bucket.
The team tried to plug the hole with volume. More than 150 marketing experiments. Most of them failed. The only paths to scale were paywalling the content or bolting on ads, and neither matched what they wanted to build. Cherkasov, as CEO, could see all of this clearly. And still he waited a year to act.
When he finally made the call, it was not gentle. The team went from 12 people to 4, and then to 2 - just Cherkasov and his co-founder, Artur Romanov, the pair having met at university and worked together on projects, papers, and startups for a decade. What followed was a strange stretch he has described as both relief and paralysis. When you clear the deck completely, you are left staring at total freedom, which is its own kind of trap.
Looking for momentum, he went to a Y Combinator Startup School event in London. A YC partner named Tom gave him a line he has repeated since like a mantra.
"Moving cautiously kills startups."
For someone who had just spent a year moving cautiously, watching a company drain out slowly, this landed with the force of a diagnosis. The response was almost violent in its speed. In two weeks he ran 54 expert interviews. He signed two product advisors. He closed three pilot contracts. He started building the MVP. The hesitation that had defined the previous chapter was replaced by a policy of relentless forward motion.
That momentum is what carried Trace into Y Combinator's Summer 2025 batch - a milestone he reached only after six earlier applications were turned down. A three-year goal, realized in a single email. His public takeaway from the whole ordeal is refreshingly free of mysticism: avoid tarpit ideas that hundreds of other founders are already chasing, and keep moving, because "any information is good information."
Trace exists because of a mismatch Cherkasov keeps pointing at. The big labs are shipping remarkably capable AI agents. Companies, meanwhile, have no clean way to decide what those agents should actually do. His framing has become the company's calling card.
In practice, Trace connects to the tools a team already lives in - Slack, Jira, Notion, email, and the rest - and builds what the company calls a context engine. It reads the internal wiki, the activity logs, the people data, and assembles a unified index of how the organization actually operates. When a new workflow comes in, Trace breaks it into steps, hands the repetitive pieces to AI agents, and routes the judgment calls to the right humans. It respects permissions, keeps a full audit trail, and lets anyone review, edit, or reroute a task.
The problem it targets is unglamorous and enormous. By the company's own estimate, most teams lose roughly 40% of their productivity to repetitive manual work - chasing updates, filling documents, coordinating people, handling support. That is the gruntwork Trace is trying to route away. More than 30 companies were using it around launch, and the ambition is bigger than a single tool. Cherkasov talks about building CompanyOS: a cognitive layer that ingests a company's internal signals and drives intelligent action across every team.
The market seems to agree the problem is real. Trace claimed the number one spot on the YC Launchpad leaderboard for the Summer 2025 cohort, a ranking driven as much by community hustle as by product. In February 2026, the company announced a $3 million seed round, with backers including Y Combinator, Zeno Ventures, Transpose Platform Management, Goodwater Capital, Formosa Capital, and WeFunder.
It would be easy to write Cherkasov up as another YC founder riding the agent wave. The more interesting thread is the one he keeps returning to himself: the year he waited too long. Most founder stories sand down the parts where the person hesitated or got it wrong. He does the opposite. He named a Product Hunt retrospective on his last company "Unwinding Our Clustr Fck." The bluntness is the point. He is a person who has decided that honesty about failure is more useful than a clean narrative.
That matters for what he is building. A tool that tells a company which work to automate and which to keep human is, in a sense, a machine for making decisions faster and with less flinching. It is a fitting thing to be built by someone who learned the cost of flinching the hard way. The manager for the AI interns is being built by a founder who spent years learning to manage his own reluctance.
He and Romanov have been at this together for over a decade, which gives Trace a rare thing among young startups: a partnership that has already survived the death of a company. Whatever Trace becomes, the two of them have proof they can walk away from something that isn't working and start again with more speed and less fear. For a company whose entire pitch is about moving work to the right place at the right time, that might be the most on-brand fact of all.