Nikita Fedorov and the problem that followed him home

Nikita Fedorov had a job, and then he had the work he did after his job. During the day, he managed engineers. In the evenings, he wrote the software he wished his team already had. The irritation was specific: testing information lived in separate systems, and getting a useful picture of product quality required too much stitching together. He kept coming back to it after hours. That second shift became Qase.

There is something appealingly unceremonious about this beginning. A company can arrive with a manifesto, a pitch deck, and a logo that cost more than its first month’s revenue. Fedorov started with a problem he recognized at work. His early ambition was close enough to touch: give the people testing software a workspace that made sense.

The second shift

At RU-Center, where he led web development, defects were reaching customers before the company caught them. He began putting testing processes in place and looking for tools. Later, while working at Avito, the classifieds business, he saw how inconvenient it was to keep manual and automated testing apart. A person could run the tests and still struggle to see the whole picture.

These were management problems with engineering consequences. Someone needed to know what had been checked, what had failed, and whether a release was ready. Moving information between tools made that judgment harder. Fedorov’s response was to start building a place where the evidence could meet.

The idea occupied evenings and nights alongside his salaried work. In 2018, he put out a free beta. A piece of software intended to solve his own working problem was now available to people whose names he did not know.

Physics, PHP, and a taste for difficult games

His route into this business ran through Omsk State University, where he studied from 2006 to 2012. He holds a master’s degree in solid-state physics. Software became his working life through PHP development, followed by responsibility for teams. By 2016 he was heading web development; in 2017, his work included leading a team responsible for billing and payment systems.

There is a considerable distance between writing a feature and being answerable for the people who ship it. Fedorov had experience on both sides. He knew the attraction of making something work, along with the less photogenic business of keeping a release from becoming a customer complaint.

Outside work, he likes side projects and video games. His favorite is Dark Souls, a game with a rather economical approach to congratulating the player. It would be fanciful to make that explain a career. It does add a pleasing detail to the portrait: the person selling better visibility into software failure enjoys a game in which failure is unusually visible.

A hundred people, then a price

Fedorov remembers spending $50 on an advertisement in a Telegram channel for the beta. The first 100 registrations followed, and some people began using the product daily. Those are modest numbers beside a funding announcement. For someone building after work, they answered a much more immediate question: did anybody else want this thing?

In 2019, Qase introduced paid plans and registered a company in Delaware. His former colleague Roman Pavlushko supplied an early investment and introductions to venture funds. The first version had proved useful enough to bring strangers in; the next stage involved asking customers to pay and finding people to help build it.

He has since said he would have charged earlier. That is a useful wrinkle in the familiar founder story. The free product attracted interest, but payment could have helped him learn the business sooner. A compliment is welcome. A subscription tells you something a compliment cannot.

Getting the evidence into one room

By August 2023, Fedorov was describing Qase through three jobs: managing tests, reporting results, and analyzing what those results meant. Manual and automated testing belonged together. Results from different kinds of tests could contribute to a shared account of the product, instead of remaining stranded with the tools that produced them.

The distinction matters because running a test and understanding a release are different tasks. A failed check needs context. A manager needs to see patterns. A developer needs to find the relevant result without conducting an archaeological expedition through the company’s software subscriptions.

That summer, Qase announced a Series A led by Chrome Capital, with participation from Finsight Ventures and S16VC. The company’s funding announcement reported more than 27 million test results processed since 2019 and named customers including Wolt, Asana, and SeatGeek. The financing would support hiring and product development. There was already plenty to build: Fedorov described a backlog of more than 500 feature requests.

“Testing pipelines have become too complex.”

Nikita Fedorov, 2023

A company cannot fit inside one developer

The same closeness to the product that helped Fedorov start Qase eventually created another problem. He could write code, speak with users, repair bugs, and sell. As the business grew, decisions about marketing, sales, and operations kept returning to him. His own account of the transition is unusually plain: doing everything himself had begun slowing things down.

He hired specialists, including leaders for operations, marketing, sales, and engineering. In describing those appointments, he acknowledged the limits of his own expertise. The change required accepting other people’s methods and opinions, along with friction of building a management team.

Geography offered a related lesson. After moving to the United States, he noticed the craft of highway advertising and began thinking differently about marketing. Qase’s distributed model let it recruit across Europe and the US. Engineering and product work could sit in one region, marketing and commercial work in another. His interest in the billboards had found its way into the organization chart.

By January 2026, Qase was describing a team spread across ten countries and more than 2,000 paying customers in over 100 countries. Fedorov’s stated preference was to hire people capable of proposing the work themselves. The founder’s job was becoming the setting of direction, with enough room for colleagues to act on it.

Teaching the tool to do more of the work

AI gave the product another set of possibilities. In 2025, Qase introduced AIDEN, tools designed to generate test cases from requirements, assess whether manual cases were ready for automation, and convert them into executable tests. The aim was familiar from the original project: reduce the work between knowing what needed checking and having useful results.

The details became more practical as the tools developed. Generated code could be exported, including working portions when a conversion was incomplete. That is an important concession to ordinary engineering life. Partial progress is still progress if someone can inspect it and carry on.

Fedorov was also trying AI development tools himself. He described building Qase Academy, a learning tool for customers, in less than a day. His colleagues promptly found a bug and suggested improvements. There is an agreeable symmetry here: the founder of a testing company produces software quickly, then the people around him do some testing.

In June 2026, the separate AIDEN name was retired in favor of Qase AI. Generation and conversion had become part of the platform’s regular workflow. The product still connected a requirement, a test, and the resulting evidence. More of the steps between them could now be performed by software.

The question of human feedback had already occupied him in a 2023 conversation with TestGuild host Joe Colantonio. They discussed generative AI, test creation, and the need for people to assess what automated tools produce. It is a useful connection to the later product work. A generated test brings its own questions: what assumptions did it inherit, what did it leave out, and can the team understand why it passed? Those questions belong in the workflow even when generating the test takes less effort than before. The person using the tool still needs a way to judge the result. A test can record an outcome; deciding whether it covers the behavior that matters requires a clear account of the request. Faster generation makes that account more useful, because the team has more output to examine.

Who checks the machines?

On October 6, 2026, Fedorov announced Tiden, a new product from the Qase team. It addressed a problem he was seeing in agent-assisted development: code, tests, and documentation could all agree with one another while drifting away from what a person had requested. An agent checking its own assumptions could produce a very tidy mistake.

Tiden keeps requirements alongside code and ties verification to those requirements. The announced workflow captures intent before an agent starts, checks changes on a branch, and leaves unresolved decisions to a person. Fedorov said his own team had been using it for months, with a small customer group trying it over the summer.

Meanwhile, the older product was dealing with a different measure of consequence. Qase’s September 2026 figures put incoming test results at eight million a day, compared with three million in March. The accompanying engineering work concerned rate limits, faster endpoints, clearer reviews, and controls over which repositories a project could access. Success had supplied plenty of unglamorous jobs.

The scale has changed since those evenings after work. The recurring interest is recognizable: give people a way to establish what their software actually does. Fedorov once needed that answer for a team he managed. Now he is building for teams whose newest colleague can write code faster than anyone can comfortably read it.