THE WORKING FILE
BRATISLAVA / Building software since 2004POST AFFILIATE PRO → LIVEAGENT → FLOWHUNTOne practical problem leads to the next

PEOPLE / SOFTWARE • BRATISLAVA, SLOVAKIA

Viktor Zeman and the problem that becomes a product

An unpaid project, an overloaded support desk, an unreliable chatbot. The co-founder of Quality Unit keeps finding his next business in the work that refuses to go smoothly.

Before the coffee came the servers. Describing his mornings in 2017, Viktor Zeman said the first thing he checked on waking was whether his company’s cloud infrastructure had suffered an outage. If something had gone wrong overnight, he often intervened himself. The coffee followed. A customer might reasonably have approved of the arrangement, even if the coffee did not.

By then, Zeman had been running Quality Unit for more than a decade. He was its co-founder and chief executive, with software used beyond his home country of Slovakia. Yet his account of the day began with a piece of operational plumbing. Somewhere, a system had to keep working. He wanted to know whether it had.

That detail helps explain a career in which one working difficulty repeatedly supplies the beginnings of another product. An unpaid customer project becomes something to sell. Supporting the buyers becomes a software problem of its own. Years later, dissatisfaction with chatbots opens another line of work. Zeman has spent much of his business life close enough to the machinery to hear it complain.

The invoice that became inventory

Zeman studied software engineering at the Slovak University of Technology in Bratislava. His education there ran from 1995 to 2000. In 2004, he co-founded Quality Unit, the company through which he would help build Post Affiliate Pro and LiveAgent. Andrej Harsani became his long-standing collaborator in that business.

The origin was less tidy than a product announcement. Recalling the beginning in a 2011 interview, Zeman described a group of self-employed people coming together. One had been working on a project for a customer who did not pay. The group turned the work into a product for sale, and found that people wanted it.

The product supported affiliate programmes. A merchant selling online could recruit partners to help with marketing, then reward them for the business they generated. Post Affiliate Pro dealt with the mechanics behind that arrangement. The abandoned payment had left the developers with something they could offer to other customers.

Distribution followed the same practical logic. Zeman said the company relied on partners rather than a large marketing effort of its own. They installed the kind of software they were selling and used it themselves. The business was learning about its product from both sides of the counter.

A RECURRING SEQUENCE
01Meet the problemInside the team’s own work
02Build the toolUse it before selling it
03Find the next needCustomers reveal the edges

An editorial reading of the path from affiliate software to support tools and AI automation.

The customers brought the next idea

Post Affiliate Pro created a second difficulty: its customers needed help. Their questions had to be answered, and the team wanted to handle those conversations in one place. Zeman later put the requirement plainly: “We needed a tool”. That modest beginning would acquire the name LiveAgent.

The desired tool would bring communication channels into one dashboard and add reporting, automation and live chat. Its starting point was the experience of supporting another product, with all the small complications that entails. Someone has already replied. Someone else needs the history. A customer should not have to reconstruct the whole conversation for each new person.

LiveAgent reached the market as live chat software in 2006. In 2011, an overhaul extended it into a multichannel help desk. Self-hosted plans moved to the cloud in 2014. Each step widened the scope of the original solution, from a conversation on a website to a larger system for organising customer service.

There were geographical changes too. LiveAgent opened an office in Kyiv in 2017 and another in New York in 2019. Its infrastructure moved from Linode to AWS in 2024. The following year, FlowHunt integrated with LiveAgent to enhance its AI capabilities. The support problem had acquired a long working life.

ONE NEED, SEVERAL REVISIONS
  1. 2004Quality Unit begins
  2. 2006LiveAgent reaches the market
  3. 2011A multichannel help desk
  4. 2014Plans move to the cloud
  5. 2025FlowHunt integration

A chief executive with unfinished code

The executive role has not erased the engineer. Zeman’s public GitHub account, under the handle vzeman, offers a useful view of the work. Its pinned projects include servers for connecting AI tools with Odoo, the Czech ARES business-information service, BiznisWeb and the European VAT-validation system VIES.

These are specific connections to systems businesses already use. An account, a company record, a tax identifier: the objects are ordinary, and making them available to another tool requires attention to detail. They fit a career built around the administrative work that sits between a business and its customers.

His public technical experiments have a similar texture. In one LinkedIn post, Zeman explained a method for adding internal links to the FlowHunt website. It involved analysing words, turning words and articles into vectors, matching them, and controlling link density so paragraphs did not become a thicket of hyperlinks.

He reported a rise in search traffic after the change, while explicitly acknowledging that he could not verify the cause with complete certainty. That caveat belongs beside the result. A working website contains enough moving parts to make confident explanations cheap and convincing evidence rather more expensive.

“Žiadnu špeciálnu prípravu nerobím.”

Viktor Zeman, 2017
“I do no special preparation.” Translation from Slovak.

The wind has no interest in the roadmap

Ask Zeman about preparing for an important meeting, and the answer is refreshingly unceremonious. In 2017 he said he had no special preparation ritual. He preferred dealing with things simply and plainly. There was no elaborate routine to reverse-engineer, no set of objects on a desk to photograph.

His life outside software contains rather more movement. He has represented Slovakia in archery and been a ballroom dancer. His public biography also lists flying small aircraft, windsurfing, kiteboarding, wingfoiling and snowkiting. The range makes the standard phrase “outside interests” seem a little underpowered.

He has resisted turning every pastime into an aid to work. Asked whether his sports helped productivity, he said he had never practised them for that reason. The distinction is appealing. Even a founder ought to be allowed a hobby without making it submit a business case.

He also described periods early at Quality Unit when sixteen-hour days and working weekends were normal. At other times, he could switch off for several weeks and kiteboard every day. His wife and children, he said, were a source of certainty and laughter. The account permits work, family and the water to exist in the same life without pretending they always fit a neat timetable.

A chatbot earns its keep

FlowHunt began with another internal need. The team wanted an AI chatbot it could trust. Existing options felt too rigid or unreliable for what it wanted to do. Experiments with automated content also exposed shortcomings, and the builder being developed began to look useful for more than one task.

The resulting platform lets people create AI tools through a visual, no-code approach. Zeman is listed as CEO and co-founder, alongside CTO and co-founder Yasha Boroumand. Its stated ambition is to give users more control over the tools they build and more time for the human parts of their work.

In his conversations about adoption, Zeman often returns to the person doing the implementing. A company can begin with one curious colleague willing to test tools and understand processes. A repetitive job provides a place to learn. The organisation can develop its experience through tasks whose results someone can examine.

His examples include descriptions, marketing content and turning existing videos into useful written material. The interest lies in work that can be repeated at scale. A small successful process can be run again, provided somebody has understood what makes its output useful in the first place.

LinkedIn Kolotočári episode artwork showing Viktor Zeman at right with the podcast hosts
Three microphones, plenty to discuss. Episode artwork from Zeman’s August 2026 conversation with LinkedIn Kolotočári. Watch the conversation ↗

First, fix the slow website

There is an agreeable complication in hearing an AI-platform founder discuss marketing. In his August 2026 podcast appearance, Zeman argued that businesses sometimes needed to fix their technical foundations before looking for an AI answer. A slow website, certificate problems and missing structured information could already be getting in the way.

He described watching AI agents fetch material from websites and discovering that some pages took an unusually long time to respond. The delay then became part of the agent’s work. A sophisticated system still had to wait for an ordinary page to load. The future, apparently, also spends time staring at the plumbing.

His interest in search now includes GEO, the effort to make a business visible within AI-generated answers. At an E-commerce Slovakia event in 2026, he discussed agentic commerce and practical steps involving structured data. The audience had shops to run; the question was how their products would be discovered as the tools of discovery changed.

At AI Boost 2025, his emphasis had included scalability and consistency. Those concerns connect the new work to the older products: handling more languages, channels and departments requires processes that can be repeated. A demonstration attracts attention once. A business needs the next run to work as well.

Starting again, with the bill attached

In August 2026, Zeman wrote candidly about bootstrapping FlowHunt. Services were bringing in multiples of the revenue earned by the product itself. Customer work supported development, while competing with it for time. His stated direction was a data layer through which AI agents could securely use a company’s knowledge, tools and work history.

The familiar difficulty had returned in a different form. The founder who had helped build established software products was again balancing immediate customer needs against a product that needed time. He could recognise the dilemma without escaping it. Experience is useful company on that road; it does not pay every invoice.

There is continuity in the ambition. Better access to what a business knows, fewer scattered conversations, less effort spent reconnecting information: these concerns run through his products. Zeman keeps returning to the work in front of him. Somewhere in its awkward details, there may be another tool worth building. First, though, someone should check that the servers are up.

Follow the work