The first version of Artem Bugara’s company existed as a boast in a chat window. He and Maksym Sugonyaka, friends since school in Ukraine and later university companions, made a Telegram chat and named it “First million.” There was no million. There was barely a product. There were two young data engineers, a shared irritation with the way information was gathered, and the useful recklessness of a name that made the destination sound almost administrative.
They started NewsCatcher in February 2020. By April, they had a beta. By June, they had put a first version on RapidAPI and found a paying customer. The motion was quick, but it did not arrive from nowhere. Their working relationship had been rehearsed for years: school in Ukraine, university, then a move to Strasbourg. There, Sugonyaka introduced Bugara to the Titanic competition on Kaggle. The two economics students began learning data science and programming. Neither had a computer-science degree. This turned out to be an inconvenience, not a verdict.
Bugara later studied econometrics at Toulouse School of Economics and led data engineers at elseco, an underwriting company working in aviation, energy and space risks. It was an unusually apt apprenticeship. Insurance turns scattered facts into judgments about exposure. NewsCatcher would eventually do a neighboring job: turn scattered public information into data a company could actually use.
A company hidden inside a small utility
The early product was easy to explain. NewsCatcher collected articles from many publishers, cleaned the results and let developers query them through an API. The first public artifacts were almost aggressively modest. One Python package wrapped Google News RSS. Another gathered normalized news from websites and filtered it by topic, language or country. Bugara described one useful tool as roughly a hundred lines of Python. Its importance was not the size of the code. It was the neatness of the solved problem.
The packages found their way onto Reddit and GitHub. During the first 60 days, the founders reported more than 200 beta sign-ups and over 800 stars. The newscatcher repository later reached roughly 3,000 stars. For a backend company, open source became a window display. Developers could test the founders’ taste before taking a sales call.
“Treat your repository as your product that you have to sell. You sell it for someone’s time.”
There is a practical generosity in that idea. A repository asks for no contract, but it still makes a demand: attention. Bugara’s standard was that the exchange should be fair. The code had to be useful on its own. In the README for newscatcher, he wrote for the junior data scientist he had once been - someone comfortable in Python but unequipped to build a complete external-data pipeline. The package was a shortcut back to the problem that person actually wanted to explore.
That instinct also shaped his writing. He published missing documentation for Google News RSS parameters, a frank account of building the beta, and lessons from the first year of bootstrapping. His prose kept the workshop dust in view: three architecture iterations, limited AWS experience, first-time-founder uncertainty. “I always prefer to learn from people who are just a few steps ahead of me,” he wrote. It is a warmer editorial principle than pretending the summit has already been conquered.
Four applications and one useful refusal
NewsCatcher’s accelerator history resists the usual clean arc. The founders applied to Y Combinator four times. They were invited to join twice. On the first acceptance, in 2021, they had already committed to TinySeed, a program designed for bootstrapped software companies, so they declined YC. At the time, NewsCatcher produced about $3,000 in monthly recurring revenue. Over the following year, the team grew and monthly revenue reached roughly $25,000.
They returned to YC in May 2022. Michael Seibel interviewed them, and NewsCatcher joined the Summer 2022 batch with about $300,000 in annual recurring revenue. By Demo Day, the founders said the figure was around $450,000. Bugara spent two days fundraising after the program and found a lead investor in Jamie Quint’s Uncommon Capital.
The numbers matter less than the change in posture. A self-serve API had attracted many kinds of customers. YC pushed the founders to decide whose problem carried enterprise weight. NewsCatcher moved toward data and risk teams, higher-value contracts and custom workflows. The company learned to qualify leads, understand the operational job and charge according to the result rather than the raw volume of API calls.
Bugara’s advice from that education is disarmingly ordinary: talk to customers and do not build features from assumptions. He admires Stripe’s founders for helping early users set up the product themselves and Airbnb for caring closely about customer needs. The common thread is manual attention before scalable elegance. Software may automate the task. Founders still have to notice which task deserves the software.
The missing 195 answers
The newest chapter returns NewsCatcher toward a self-serve product, but the question has changed. Conventional search is designed to rank a few useful pages. That works when someone needs one answer. It is much less comforting when a risk analyst asks for every regulatory action in an industry, every cybersecurity incident in a week, or every facility expansion announced across hundreds of regional publications.
If 200 valid events exist and a search system surfaces five, the result may look polished while achieving 2.5 percent recall. A language model can summarize the five beautifully. It cannot reason over the 195 it never saw. Bugara and Sugonyaka now describe NewsCatcher’s CatchAll as a recall-first web search API for these “long-list” questions.
The system pulls a large set of candidate pages, checks which ones match, extracts user-defined fields and deduplicates the events. The output is a dataset rather than a row of blue links. A query can become a monitor, which makes the search useful for repeated work: supplier disruptions, lawsuits, product launches, funding rounds, sanctions or local incidents that gather consequence quietly.
This is the same founder story at a different scale. The old open-source package helped a developer avoid assembling a news pipeline. The new product tries to keep an AI agent from browsing page by page until its context window or patience expires. In both cases, the founders are packaging the tiresome middle: collecting, checking, cleaning and presenting information in a form the next person can use.
The small local report may carry more operational value than the headline everyone has already read.
Consistency, without the costume
Bugara is unusually willing to preserve the awkward parts of the record. He calls himself a first-time founder. He notes the lack of a computer-science degree. He has spoken about starting outside the familiar pipeline of famous universities and large technology companies. His conclusion is not self-mythology. It is consistency.
That word fits the evidence. He and Sugonyaka learned together, shipped together, reapplied together and changed the company’s customer more than once without abandoning its central material: public information converted into usable data. Their friendship has lasted nearly two decades. The product has moved from RSS and news crawling to enterprise enrichment and event-centric web retrieval. The line between those points is crooked, but visible.
There are lighter clues, too. On a podcast, Bugara told listeners that if they were ever in Warsaw, they should reach out and he would be happy to get coffee. He told another interviewer that he recharges with his wife and dog, and that writing before meetings had made a disproportionate difference. These are small admissions from someone building systems for enormous quantities of information: a preference for preparation, companionship and the occasional conversation that has no API endpoint.
The “First million” chat was a joke with ambition inside it. Six years later, Bugara’s more interesting unit is not money but coverage: the result on page 196, the obscure source, the event no ranking system thought glamorous enough to show. Search has spent decades learning what to put first. He is working on what it should not leave behind.
His stated ambition for NewsCatcher is broad access to real-time information, so that companies of different sizes can make quicker, better-grounded decisions. CatchAll gives that old ambition a sharper technical shape. Equal access means little if the underlying retrieval is patchy. A trustworthy feed must reach beyond familiar publications, across languages and regions, then leave a trail that an analyst can inspect. For Bugara, completeness is becoming less a search metric than a promise about who gets to notice what is happening.