When Leon Iwanowitsch was 12, he launched his first business selling honey. Since then, he’s finished his studies at St. Gallen, worked at Robin Capital as a Founder’s Associate, joined Y Combinator, and is now building Ontora.

Leon’s idea started with a problem he experienced himself at his last job in AI consulting. AI made building solutions fast. Figuring out how a company really worked- the workflows, bottlenecks, and know-how that never make it into any document- still took months of interviews. “The real challenge lay in gathering operational knowledge from employees,” he says.

Ontora aims to bridge that gap. Its AI agents interview employees across the company to understand how work actually gets done, surface bottlenecks, and hand that operational context to the tools that actually build solutions.

For Leon, this is the next step in a founder’s journey shaped by luck, experimentation, and learning to handle uncertainty. Join us as he shares what he’s learned along the way.

About Ontora

Ontora builds AI agents that interview employees to understand how work actually gets done — the workflows, bottlenecks, and know-how that never make it into any document — then hands that operational context to the tools that actually build solutions. Backed by Y Combinator and Robin Capital, Ontora was founded by Leon Iwanowitsch, Maximilian Arnold and David Korn, who met through the START Summit ecosystem before joining forces to build the company.

From supporting founders to becoming one

Entrepreneurship isn’t new for Leon. “I've always had a strong entrepreneurial spirit,” he says, remembering that it started when he sold honey at age 12.

Working at Robin Capital gave Leon a much clearer picture of what it really takes to build a company. Being close to founders showed him just how complex and uncertain entrepreneurship can be.

It also changed how Leon thought about investing. He noticed that many of the investors he admired had once been founders themselves. “The most successful VCs and LPs are often former founders themselves,” he says, “as they truly grasp the complexities and uncertainties that come with being a founder.” Eventually, Leon wanted to face those challenges himself.

“I've always had a strong entrepreneurial spirit, from my first ‘small startup’ at age 12 selling honey, to now, after finishing YC.”

Creating your own luck

Leon describes his journey to Ontora as a great example of the “butterfly effect.”

At St. Gallen and START Global, he met entrepreneurs who repeatedly spoke about the role luck had played in their success. Looking back, he can trace his own journey through a similar series of connections: attending the University of St. Gallen led to START, START led to Robin, and that connection ultimately helped him land the role where he encountered the problem that he would go on to solve with Ontora.

This experience changed how Leon thinks about luck. You can’t control when opportunities show up, but you can put yourself in more situations where something unexpected could happen.

That means meeting new people, trying new things, and having conversations even if you don’t know where they’ll lead. It’s one reason Leon likes San Francisco. Here, a random meeting could connect you with a customer, investor, or idea that changes your company’s path.

“You cannot influence luck per se, but you can increase your surface for luck.”

Finding the missing context

Leon came up with the idea for Ontora while working at an AI consultancy for enterprise clients.

He noticed that while modern AI could generate solutions quickly, understanding the real problem was still a slow, manual process. Teams often spent months interviewing employees, piecing together what they learned, and finding bottlenecks before starting on a solution, discovery hadn’t sped up when development did. “The real challenge lay in gathering operational knowledge from employees through interviews,” Leon says.

Leon began to wonder whether the information-gathering process could be automated as well. If companies could capture and organize what their employees know, AI would have a much clearer view of the organization. This was the starting point for Ontora.

“AI made building solutions fast. Knowing what to build is the bottleneck. We knew there had to be a way to automate the interviews, structure what people actually know, and hand that context to the tools that build.”

Giving AI access to how companies really work

Ontora runs those interviews with AI agents at company scale — so leaders get a map of how work actually gets done before they decide what to automate.

A company might start with an objective such as identifying opportunities to automate processes within its sales organization. Ontora gathers information about the business, identifies the relevant employees, and invites them to speak with its AI agent through channels such as email, Slack, or Teams.

Unlike a traditional survey, the conversation isn't confined to a predetermined list of questions. The agent responds dynamically, following up on individual answers and digging deeper where necessary.

Once those conversations are complete, Ontora can synthesize insights across the company, surface patterns and feed the resulting information into a company's wider AI infrastructure.

What used to take weeks of one-by-one interviews can now run in parallel across the company. Internal teams move faster on the solution work, and consultancies and transformation partners can use the same motion to scale discovery without turning Ontora into a consulting substitute.

“Just as humans would do it, that's exactly the way our AI agent does it.”

Turning how work gets done into context for AI

Ontora is built around tacit knowledge, the information employees have in their heads that rarely ends up in databases, documents, or formal processes.

Employees spend much of the workday sharing this kind of knowledge. Teams meet, managers talk to employees, and departments exchange information because companies need to know what’s happening beyond what their systems show.

“In essence, we help companies access how work actually gets done, the tacit operational knowledge AI needs, at scale,” Leon says.

Right now Ontora is focused on AI transformation buyers: before a company can see where AI will help most, it needs to understand what employees really do, how tasks fit together, and where bottlenecks exist, then hand that living context to the tools and agents that build solutions. This creates a foundation for more tailored AI solutions.

“Most companies are still in the early stages of AI transformation."

Finding the right co-founders

Leon met his co-founders years before Ontora was even an idea.

While at St. Gallen, Leon organized the START Summit, which his future co-founders attended. They stayed in touch, but it wasn’t until a later trip to Stockholm that they realized they were all tackling the same problem from different angles.

“We realized we were all addressing the same problem from different perspectives,” Leon says.

They decided to join forces. He describes the resulting founding team as a Venn diagram: each person brings distinct technical and non-technical strengths, while overlapping on what matters most: vision, working style, and motivation.

“Each of us brings unique technical and non-technical skills, while also sharing a strong overlap in our vision, working style, and core motivation.”

The YC lesson: Just try it out

Ontora entered Y Combinator at an unusually early stage. The company itself was formed during the first weeks of the program, leaving the founders to build the product while simultaneously finding their first customers.

Their initial instinct was to approach the problem analytically: develop hypotheses, identify potential contacts, and reason through the best strategy.

Their YC partner Tom Blomfield suggested a simpler approach: “You don't know anything yet, so you just have to try it.” For Leon, this showed the limits of thinking through every early-stage problem. “When you lack data, you can't make informed decisions,” he says. Instead, you need to act, see what happens, and use that to decide your next step.

It was a shift from the academic mindset Leon was used to. Building a startup was not about finding the way forward before starting. It was about learning through experimentation.

“Try something that you have an intuition for, gather data with that, reflect on that, and try again.”

Building for better AI models

When the technology behind your product gets better every few weeks, each new model release might seem like a threat. But Leon sees it differently.

He recalls a conversation with Sam Altman at YC about a useful test for AI companies: if tomorrow's models become significantly better, does your product improve or become obsolete?

“For us, the answer is clear,” Leon says. “We welcome better models because they can reason more effectively with the information provided.”

Ontora isn’t trying to build a smarter foundational model. Instead, it’s betting that even the best models need good information about how a company really works. As these models get better at reasoning, giving them the right information becomes even more valuable.

“The focus isn't solely on how intelligent a model is but on the context we supply.”

Quickfire round

A tool you can't live without?

“No one wants a tool — everyone wants the outcome. A tool only earns its place if it clearly delivers that.”

A founder you admire?

“Someone I really admire is Piet Terheyden, co-founder of Pool. He’s very down-to-earth and very modest. He has great achievements, but he doesn’t talk about them much.”

How do you recharge?

“If I have a recharge ritual, it’s playing piano. Music gives me a way to process whatever is happening around me. Whatever emotion I have, I can have a piece in mind to play. Music is just another type of communication.”

Advice that has stuck:

“Embrace the pain” — from Robin. It's about trying to perceive it as a moment you learn and grow instead of dwelling in the pain itself.

Handing living context to AI tools

In the end, Leon wants Ontora to do much more than automate interviews.

“Organizations conduct hundreds or even thousands of internal meetings daily, primarily to share context and transfer knowledge,” he says. Much of what a company needs to know already exists somewhere inside the organization, but remains fragmented across people, teams and conversations.

Leon sees Ontora as the way companies continuously capture how work gets done and feed that context into their AI tools and agents, so those systems reason over real operating reality, not just documents. (Think of it as living operational context for AI, not a static knowledge base.)

For AI, this could mean a much deeper understanding of the company it’s helping. And for leaders, it could offer something just as valuable: a clearer picture of what’s really happening inside their business.

Interested in what Ontora could uncover inside your organization? Reach out to Leon on LinkedIn or head to Ontora’s website to learn more.