The book came recommended by a friend. Artur Saudabayev thought Yiannis Kiachopoulos should read Jared Diamond’s Guns, Germs and Steel. Kiachopoulos did what an interested reader does: he read, took notes and tried to keep hold of the argument. Then the notes disappointed him. He could follow the relationships on the page; capturing the whole arrangement was another matter.
Saudabayev had encountered the same frustration. The two computer scientists began exploring whether algorithms could turn text into diagrams of cause and effect. Their shared reading problem became an early impetus for Causaly, the company they would build together. It is an unusually domestic beginning for an enterprise software business. Before the investment rounds and institutional customers, there was a reader with an inadequate notebook.
That beginning makes Kiachopoulos easier to understand. He is interested in the distance between encountering information and being able to use it. His work keeps returning to the inconvenient stretch in between.
Japanese, code, and the enterprise
Kiachopoulos’s education ranges across Japanese language studies, computer science and business administration. He holds a master’s in computer science from FAU Erlangen-Nuremberg, completed with distinction, and an MBA from the Hong Kong University of Science and Technology. His MBA studies included an exchange at London Business School.
It is a pleasingly untidy collection of subjects. A founder biography ordinarily tries to make every qualification look like preparation for the final job. Here, the variety is interesting enough on its own. Language, computing and management each appear in the record; there is no need to pretend that the company was inevitable.
His earlier professional life involved the systems that large organizations depend on. At Accenture in Germany, he worked in IT consulting, including security and infrastructure management. Later, as a principal at Detecon International, he advised telecommunications clients on mergers, acquisitions and organizational transformation across Europe and Asia. These were established enterprises with existing people, processes and obligations.
That background gives his later writing about AI adoption a practical context. He asks about responsibilities, review and the way work passes between colleagues. Those questions belong as much to management as to software engineering.
A prototype before a pitch
The founders met in California at Singularity University’s summer program. Kiachopoulos became Causaly’s chief executive; Saudabayev became its chief technology officer. The company dates its founding to 2018. Their complementary responsibilities would matter when the project acquired the usual complications of a business.
In an early experiment, they ran a prototype across several thousand Wikipedia articles. The question was whether a computer could extract relationships from prose and make those relationships useful. They began with the technical possibility, then considered where to apply it.
That sequence is revealing. A sentence contains more than a collection of searchable words. It can make a claim, qualify another claim or explain why something happened. A reader follows those connections almost without noticing the work involved. Asking a machine to expose them produces a considerably longer engineering assignment.
Kiachopoulos has described the next difficulty in customer terms. Understanding a user’s daily work required deliberate attention. The team’s scientists spoke frequently with users, treating their problems as something to work through together. The challenge extended beyond getting the technology to function: it had to fit a task someone actually needed to finish.
Someone has to keep the company running
“We loved bootstrapping,” Kiachopoulos recalled of the early period. Self-financing gave the founders room to experiment. External investment followed as they learned more from a customer pilot and wanted to deliver the product faster.
By late 2019, the company had a ten-person team across London and Athens. Kiachopoulos spoke plainly about the cost of fundraising in attention. Time spent securing money could leave less time for customers, colleagues and the product itself. For a founder, those demands arrive on the same calendar.
He credited Saudabayev with giving him confidence that the company would keep moving while he concentrated on the financing. It is a useful detail about their partnership. Co-founder trust becomes concrete when one person can spend a day elsewhere without imagining that every unresolved question will follow them into the meeting.
He also chose investment partners for relevant experience. Pentech’s Marc Moens brought knowledge of natural language processing and its commercialization in business software. In that account, the identity of the investor mattered alongside the money. Expertise could be useful after the announcement had stopped attracting attention.
The checks got larger.
The subsequent financing shows the scale of the enterprise he was assembling. Causaly announced $4.8 million led by Pentech in November 2019, then $17 million led by Index Ventures in May 2021. In July 2023, ICONIQ Growth led a $60 million Series B. The company put its cumulative funding at $93 million at that point.
Datadog co-founder Olivier Pomel and former chief executive Alex Gorsky were among the individual participants in the 2023 round. Carlos Gonzalez-Cadenas at Index and Caroline Xie at ICONIQ became part of the company’s investment story. Kiachopoulos was building relationships across enterprise software and the organizations buying it.
The amounts establish a dated record of investor backing. They also make the change in his job visible: the reader experimenting with algorithms was now responsible for a business with a growing network of employees, customers and financial partners.
The team on the other side of the map
London is Kiachopoulos’s public base. Greece has been part of Causaly’s engineering story since its early years, and the company also has an office in New York. Its geography reflects the work of assembling a team across more than one market.
At the Delphi Economic Forum in 2021, Kiachopoulos discussed a Greek team of roughly twenty engineers and scientists. He described a startup ecosystem that had changed markedly since 2012. As more companies emerged, finding and retaining people became more competitive.
It was a founder’s observation about success creating its own difficulties. A stronger startup community means more possible employers for a skilled engineer. A growing company has to give people reasons to join and reasons to stay.
Causaly entered Endeavor Greece’s Scale Up program in 2021. Kiachopoulos’s public involvement places him among founders discussing what the wider ecosystem needs, alongside the demands of his own company. The talent conversation is particularly tangible for someone whose business depends on engineers and specialist colleagues working together.
“Everything needs to be inspectable.”
Yiannis Kiachopoulos, June 2026
There is also a small, cheerful clue to his interests in his Wellfound profile: computational linguistics and graph databases appear in his own description of the work. These are specific enthusiasms, close to the mechanisms behind the original reading experiment. They give some texture to a title that could otherwise sound purely commercial. The chief executive still presents himself through the technical problems the company began with.
His public role has widened with the business. At Causaly’s October 2023 gathering in New York, the program put him in an opening address, a fireside conversation with Alex Gorsky and a presentation on the company’s future direction. Explaining the work, discussing it with experienced executives and showing where the product might go all fell within the same day. A founder’s calendar can become a fairly accurate description of the job.
An answer you can take apart
Inspectability is a recurring concern in Kiachopoulos’s public explanations of AI. A user should be able to trace an answer, examine its evidence and challenge the reasoning. The confidence of the writing cannot do that job for them.
His July 2026 essay on AI architecture widens the frame to the system surrounding a model. Planning, memory, tools, evaluation and human review have to work together. One component can perform adequately while another loses the context that made its output meaningful.
For a chief executive with an enterprise consulting background, this is familiar territory in a new technical form. A capable component still has to belong to a workable organization. Passing information between stages can be as consequential as producing it in the first place.
In August, he extended the argument to adoption inside a business. Connecting a model to proprietary data leaves questions about methods, checks, ownership and decision rights. Individual productivity does not automatically change the way a department works. The organization has to decide what it will accept, who will review it and how corrections travel back through the process.
What happens after the answer?
- 01Keep the evidenceMake the underlying material available.
- 02Expose the reasoningShow assumptions and unresolved questions.
- 03Give someone the reviewDefine who can accept or challenge the result.
The work has to survive the handoff
By May 2026, Causaly’s Scientific Workflows offering was turning established expert methods into processes that AI agents could execute repeatedly. The intended value included consistency across teams, with checkpoints and evidence trails available for review. It brought Kiachopoulos’s management questions into the product itself.
September brought an integration with Microsoft 365 Copilot and Teams. Causaly’s research capabilities could be invoked within tools enterprise employees already used. The announcement concerned access as well as capability: a useful system has a better chance of being used when it appears inside the working day.
His October 20 essay returns to a basic question: what assignment should the system complete? He argues for defining the expected output and its standard of completion, then examining the result. Unfinished items must remain visible. An agent should know when to involve a person. Presentation has to preserve the meaning of the underlying work.
These are exacting, rather unglamorous requirements. They suit the original frustration. A reader can reach the last page and still lack the notes he needs. A machine can return an answer and still leave the assignment incomplete. Kiachopoulos has spent the intervening years building a business around that remaining work. The notebook, it turns out, had given him plenty to do.