Thanos Papangelis would rather build it himself

Thanos Papangelis has a city to run. It lives in a browser tab, requires no installation, and has the considerable advantage of being one he built himself. After years helping organizations train their employees, the Epignosis co-founder has returned to a familiar occupation: making something because he wants to use it. Roads, buildings and the consequences of questionable planning are included.

His city simulation is called Polis. He began building it in his spare time with help from Claude, the AI assistant. By July 2026, the experiment had acquired company: a chess game and a three-dimensional strategy game, gathered under PixelAgora. A founder who once worried about reaching customers around the world was now removing the small irritations between a player and an evening’s entertainment. No account. No payment. No advertisements.

The requirement for admission to this collection was personal: he had to enjoy playing the game. There is a pleasing freedom in that standard. Software usually arrives accompanied by a proposal for how it will occupy your wallet. Papangelis’s games ask for a little time, which is quite enough to ask of anyone.

“I don’t have a business model.”

Thanos Papangelis, on his browser games, July 2026

A city in a browser, a founder at the keyboard

Polis has an older city behind it. As a child, Papangelis spent hundreds of hours playing SimCity. He credits those hours with teaching him about resources, systems and the inconvenient distance between a decision and its consequences. An apparently harmless budget cut can wait patiently before making its objections known.

He sees a resemblance in companies. Neglect product quality and the damage may arrive much later, when the original choice has disappeared beneath newer decisions. A dashboard records what has already happened; understanding a system requires thinking about what might happen next. For a founder, a city simulation offers a fairly economical way to acquire some expensive instincts.

Even the name carries a working relationship into the game. Papangelis credits his co-founder with the idea behind Polis, recalling an internal message about community and private self-interest. The connection is easy to understand: a city struggles when every district pursues its own advantage. A company has similar difficulties when each department wins its private competition.

An email, an examination, a change of plan

Papangelis was born in Edessa in northern Greece. His father was an accountant; his mother worked in banking. Entrepreneurship was not the family trade. Computers supplied a different possibility. An encounter with one at a friend’s home caught his attention, and books and self-taught programming followed. Before software became a profession, it was something he wanted to investigate.

Formal study took him to the University of Patras: computer engineering, a master’s in software quality, and eventually a doctorate in computer science. It is a useful education for someone who would spend years asking why a piece of software needed to be so difficult. The attention to quality preceded the company’s eventual appetite for simplicity.

In 2001, an email from Dimitris Tsingos reached an alumni mailing list. Tsingos was looking for collaborators. Papangelis responded, and a partnership began. Their early ambition concerned Greek students preparing for university admission exams. The business that followed would travel a long way from that examination room.

The project began in 2003. By 2007, after military service, Papangelis committed himself to it full-time. Greece offered a limited market for what they were making, so the founders turned outward and adopted an open-source approach. The early idea had supplied a beginning; finding a workable audience required a willingness to revise it.

Make it simple enough to buy

TalentLMS arrived in 2012 with a proposition that was easy to appreciate: workplace training software should be straightforward to start using. The product was offered as a service, with a self-service approach and affordable entry. Buyers could begin without waiting for a grand organizational ceremony. In enterprise software, dispensing with a ceremony can amount to a small act of courtesy.

Papangelis’s preference went deeper than the purchase screen. He wanted customers to understand the product without needing someone to explain it. That places a burden on the maker: instructions, setup and everyday use must do work that a conversation might otherwise conceal. Simplicity becomes an engineering obligation.

The product family grew to address different working lives. eFront serves organizations needing more customization; TalentCards puts training on mobile devices; TalentHR handles people-management tasks. Papangelis helped initiate and guide these products alongside TalentLMS. The common question is practical: how does the tool fit the person who has to use it?

A worker away from a desk needs a different route into learning from someone sitting at a computer all day. That distinction helps explain why the portfolio contains several products. A clear idea about access can produce different forms, each shaped by the circumstances of the learner.

The programmer learns to let go

Building the company brought a less agreeable design problem: the founder himself. In the early team, Papangelis moved across sales, marketing, product and customer success. He later recognized how easily that arrangement could make him a bottleneck. Having a useful answer to everything is a splendid way to ensure that everything waits for you.

He describes himself as an introvert. Some parts of being CEO did not come naturally, and handing technical leadership to other people was difficult. The programmer still wanted the keyboard. With support from his co-founder, investors and board, he learned to let others take responsibility for technology while he concentrated on leading the business.

The distinction matters in a company devoted to learning. Its founder had to acquire a skill for which programming offered limited preparation: allowing capable people to do work he cared about without doing all of it himself. A growing organization needs room for other people’s judgment.

At the end of 2023, Papangelis stepped down as CEO, remaining on the board and becoming a product and innovation advisor. Nikhil Arora succeeded him. The announcement described learning software used by 22 million people in more than 160 countries. Those figures belong to that moment; they also show how far the original Greek project had travelled.

The leadership arrangement subsequently changed again. Tsingos returned as CEO in 2025. Today, Papangelis is listed as co-founder, Chief Research Officer and board member. Research gives his technical curiosity an official address, while the independent projects provide another place to put it to work.

An investor who asks for proof

Company building also carried Papangelis to the other side of the investment table. In July 2026, he wrote that he had invested in more than 20 startups, as well as backing funds. His own preference as an angel had become more specific: founders should demonstrate something real, then use a meaningful round to reach a destination that can stand without another investor’s permission.

He distinguishes that choice from his role as a fund investor. Larger funds can have the reserves and influence to support businesses through repeated rounds. A small direct investor has different powers. His argument concerns the fit between the money and the business, including what happens to ownership as more money arrives.

It is a less theatrical view of entrepreneurship than the familiar celebration of the next cheque. Customers, a team that can deliver, and a viable business are harder to photograph. They are also what remains when the announcement has finished its social rounds.

Getting the thought onto the page

The same tools helping Papangelis make games have changed his writing. In a January 2026 essay, he described keeping thoughts to himself because he could not always express them clearly. Introversion, in his account, left a gap between having an idea and getting it onto the page.

AI gave him a way to work through that gap. Fragments, rough paragraphs and voice notes could become material to examine and revise. He retained responsibility for the opinion and the result. The useful part was getting closer to the thought he had been trying to express.

That account gives Downstream, his personal website, a particular interest. These are the observations of someone testing tools against his own habits. Writing becomes another product problem: identify the obstruction, try something, and judge whether it helps. The audience gets to read what previously might have remained an unfinished thought.

The pleasure of making, the duty of checking

By September, his experiments included Cognitive Bias Explorer, an interactive map of 188 biases. It offers examples, exercises and ways to examine the shortcuts people use when making decisions. The project follows a question running through his recent work: when producing an answer becomes easy, what happens to the effort of deciding whether to believe it?

His October essay about AI and authority brings the question home. He acknowledges checking some answers less carefully as the systems become familiar. There is a revealing modesty in the admission. Mapping a bias does not confer immunity; the person drawing the map still has to find his way.

The games, the writing and the research role now sit beside one another. All give Papangelis opportunities to build, watch what happens and reconsider. The practical curiosity that began with programming has more outlets, and making something small no longer requires assembling a company around it.

For someone who helped spend years taking learning software around the world, a game in a browser is a pleasantly immediate reward. Open it, make a decision, see what follows. The city may object to your budget. At least it will give you something specific to fix.