Victor Riparbelli and the art of being useful

Before Victor Riparbelli was selling businesses a new way to make video, he was helping them get onto the internet. At 17, he could set up an online shop in a few hours and charge for the work. The customer saw an expensive technical obstacle. He saw something he could assemble. It was a modest transaction, and a useful education in the distance between what technology can do and what people can do with it.

He also had another career in mind. “I always wanted to be a musician,” he has said. Electronic music remains a spare-time pursuit. The CEO of Synthesia still makes tracks, a pleasingly human detail for someone whose working life involves making digital people speak. Even an avatar business leaves room for someone who wants to make music.

The shop-building teenager and the music enthusiast make an instructive pair. One wanted to remove a practical obstacle; the other wanted to create. Riparbelli’s career has kept returning to both impulses, sometimes with a considerable argument between them.

The paper that changed the plan

Riparbelli grew up in Denmark and studied at the IT University of Copenhagen. A semester at Stanford exposed him to people who treated ambitious company ideas as reasonable subjects for conversation. He considered the United States, then chose London, where he could pursue his interest in virtual reality without first taking a job at a large company to secure a visa.

His curiosity about emerging technology was already taking him through virtual and augmented reality. Those fields suggested new ways to create and experience media. Video offered another possibility: apply new computing techniques to a medium people already used, rather than wait for everyone to acquire new equipment and habits.

In London, he met computer-vision researcher Matthias Niessner. The Face2Face research showed facial reenactment in video. To Riparbelli, it suggested a future in which the machinery of production could become software. In 2017, he co-founded Synthesia with Niessner, fellow Danish entrepreneur Steffen Tjerrild and computer-vision professor Lourdes Agapito. The founding group brought commercial instincts and academic research into the same venture.

Getting other people to finance that venture was difficult. His account of the early fundraising includes more than 100 investor rejections, followed by a cold email to Mark Cuban that led to a $1 million investment. It is an appealing startup anecdote, although the less cinematic part matters too: the company still had to decide what it could build and who would pay for it.

“utility over novelty”

Victor Riparbelli’s product principle

The audience without a film crew

Initially, Synthesia worked on translating and dubbing existing footage, changing voices and mouth movements for different languages. Its prospective buyers were film studios and advertising agencies. The technology could attract attention. The commercial problem was that those customers already knew how to produce video, and dubbing addressed one part of a much longer process.

Riparbelli came to see a different audience: employees who needed a video but lacked the budget, equipment or production skills. They were making documents and presentations because those were the tools within reach. An imperfect new method could still expand what they were able to do. The relevant comparison became the work they could actually finish.

A training manager does not begin the day with a screenplay and a cinematographer. Someone needs to explain a process, introduce a product or help a new colleague get started. The production question can swallow the communication question. Synthesia’s shift toward avatar-led business videos put the latter back within reach of the person with the information.

Riparbelli has described the early task as finding a first step the team could both build and sell. The technology was not ready to deliver every ambition. Talking to prospective users helped turn a broad vision into a narrower product. That required a certain willingness to be disappointed by the original plan, then interested in what the disappointment revealed.

An edit button earns its keep

There is a personal logic to the product. When Riparbelli wants to learn something, he has described starting with YouTube or a podcast and getting to the book later. His business addresses the mismatch between those habits and workplaces where knowledge arrives as another document. A familiar voice explaining something has an appeal that a file attachment must work harder to earn.

Still, making a talking face is only part of making usable software. In conversations about Synthesia, he returns to collaboration, publishing, permissions and the surrounding workflow. These are features that rarely make anyone interrupt lunch to watch a demo. They are also the features a team encounters after the first demonstration has worn off.

He has acknowledged mistakes in both directions: pursuing new capabilities too eagerly and waiting too long. That is a more useful description of running an AI company than a smooth retrospective in which every decision was inevitable. Technical advances arrive unevenly. A product roadmap has to meet customers who still have deadlines during the uneven bits.

Control is another recurring concern. In his discussion with developer-tool founder Guy Podjarny, Riparbelli explored editable video, personalization and the importance of human craft. A generated result becomes more valuable when a person can change it deliberately. The work continues after the computer produces the first version. Anyone who has ever revised a presentation should recognize the situation.

He also expects new entertainment formats to emerge from younger creators working outside established studios. In that view, easier production would give people with ideas and limited budgets more room to experiment. The attraction is familiar to anyone who has tried making music on a computer: tools can make a creative attempt possible long before an institution would pay for it. What those creators choose to make remains their decision.

His own use of AI can be wonderfully ordinary. He asks for naming suggestions, uses assistants for research and reshapes written material into scripts. He rarely takes a suggested product name exactly as offered. The machine supplies possibilities; he retains the decision. It is a small example of the control he wants users to have over larger creative tasks.

A London address, a larger argument

By April 2025, Synthesia announced that it had passed $100 million in annual recurring revenue and received investment from Adobe Ventures. That recurring-revenue figure describes the business’s subscription run rate. It marked a commercial milestone for a company whose early difficulty had been finding a repeatable use for an unusual technology.

The London headquarters opening that summer put the business in a different kind of frame. Riparbelli was joined by Technology Secretary Peter Kyle, Mayor Sadiq Khan and AI researcher Fei-Fei Li for the opening of a 20,000-square-foot workspace. Building a company eventually involves a surprising amount of floor space for an undertaking that began with the possibilities of a laptop.

In January 2026, a $200 million Series E led by GV valued Synthesia at $4 billion. The financing also included plans for an employee secondary sale, allowing long-serving staff to sell some shares. A company’s growth changes the stakes for the people inside it as well as the founder whose name appears in the announcements.

Riparbelli’s ambitions extend to the European environment around the company. He helped establish Project Europe, supporting younger founders. Building from London gives his arguments a practical setting: he wants European companies to compete globally, and he has been willing to criticize the conditions in which they are expected to grow.

His public writing also offers a longer record of the questions occupying him. Posts on synthetic media and its ethics sit alongside earlier writing about startup growth and the employee experience. The subjects connect: new tools still enter organizations made of people, with habits and expectations that a technical breakthrough does not automatically rearrange.

When the lesson answers back

In his TEDAI Vienna talk, Riparbelli pushed the communication argument considerably further, proposing that AI could help audio and video displace text as the primary way people exchange information. It is a forecast, and a provocative one. The more immediate question is easier to test: when would hearing, seeing or practising help someone understand?

On October 1, 2026, Synthesia launched Sessions. The platform offers conversations with interactive avatars, including roleplays and surveys. The change introduces a new demand on the product. A prerecorded presenter can deliver an explanation; a conversational system has to respond to the person who interrupts, misunderstands or asks for another example.

On October 22, Synthesia announced a partnership with recruitment company Hays following a Roleplay Sessions pilot. The reported results included voluntary engagement above 70% and a 29% improvement in measured core sales skills among the participating cohort. Those are pilot findings. Their significance will depend on what happens as people use the tool over time and in everyday work.

Riparbelli’s stated ambition for the new funding is to help employees adapt, learn and make better use of AI themselves. The continuity with his early work is striking. There is still someone facing a task that seems too expensive or difficult to arrange. The founder’s job is to make the next attempt possible.

The teenage shop builder could complete a useful piece of work in an afternoon. The video company has taken years of research, changed plans and customer conversations. Underneath the difference in scale sits a familiar test: has the person on the other side acquired an ability they needed? For a founder with music still in his life, there is a satisfying restraint in that question. Let the audience decide whether to play it again.