Jérémie Berrebi does not begin his morning by opening a startup portfolio.

He studies Talmud.

The afternoon can move from a product to a company, from a technical question to an investment decision. Work may resume in the evening. He travels little, avoids cocktail events where everyone meets and almost nobody really talks, and lives in Bnei Brak, Israel, in a family with eighteen children.

The profile sounds assembled from incompatible pieces: French internet pioneer, entrepreneur, hyperactive investor, corporate adviser, religious student, father and grandfather, then once again a maker sitting in front of AI tools.

There is a surprisingly consistent line through all of it.

Berrebi likes the beginning of things.

The moment before a product has a category. Before a market has decided how work is supposed to be done. Before processes, meetings and organization charts arrive to help a small team stop moving quite so quickly.

The internet before the profession existed

In 1994, the web was still new enough that “working on the internet” did not describe much of a profession. Berrebi managed online communities on CompuServe, including around CNN, then wrote about technology for ZDNet France.

He was barely twenty when he moved to the other side of the screen.

Net2one emerged in the late 1990s around an idea that now sounds ordinary: aggregate news, let a user choose keywords, then send a personalized selection every morning.

Google News did not exist. RSS had not yet turned syndication into routine plumbing. The team had to build its own extraction tools, understand site templates, retrieve stories, index them, assemble hundreds of thousands of newsletters and deliver them without setting the servers or spam filters on fire.

Net2one reached several hundred thousand users. Berrebi also became one of the public faces of France's “new economy.” Newspapers, television and political events invited the very young founder to explain what the internet was becoming. Meanwhile, the company grew from a small group to dozens of people in a matter of months.

That expansion contained an early lesson that would recur throughout his career: a company can become bigger much faster than it becomes better.

The dot-com bubble burst in 2000. Capital stopped behaving like a default fuel source. Net2one reduced its team, gradually shifted the free product toward paid services, found professional customers and was eventually sold in 2004 to Presse Plus, then part of the TNS/Kantar world.

Berrebi sold his Paris apartment the same day. Soon afterward, he moved to Israel with his family.

He was not yet thirty. He had already lived through a complete cycle that many founders never see: product, hypergrowth, visibility, crisis, downsizing, business model and exit.

Zlio and the pleasure of arriving too early

Zlio came next.

The original idea was to let anyone recommend products and earn a commission. It evolved into something more visual: create an online store in minutes even if you had no inventory and nothing of your own to sell.

Zlio aggregated merchant catalogs. A user assembled a themed shop. When a visitor wanted to buy something, they were redirected to the actual merchant and the affiliate relationship paid for the recommendation.

Dropshipping, creator commerce, niche storefronts, mainstream affiliate marketing: much of today's vocabulary was not there yet, but several of the mechanics already were.

Hundreds of thousands of shops were created. Mashable recognized the service. Google became interested enough to discuss acquiring the company and hiring Berrebi. The deal did not close.

A few months later, Zlio's visibility in Google collapsed. A platform whose distribution depended heavily on search abruptly discovered that it did not own the pipe feeding it.

Zlio survived several more years before closing in 2011.

From a distance, the episode looks like a failure following a missed exit. Inside Berrebi's larger story, it installed a more durable instinct: never confuse visible growth with ownership of the system producing that growth.

Kima turns investing into cadence

In 2010, Jérémie Berrebi and Xavier Niel created Kima Ventures.

Its first public objective looked almost absurd at the time: invest in 100 startups within two years, globally, with tickets ranging from €5,000 to €150,000.

The idea was not merely to create another seed fund. It was to compress the time between discovering a company and making a decision. Where a conventional fund might multiply meetings, Kima wanted to look at an enormous number of projects, decide quickly, then do it again the following week.

The pace reached one or two new investments per week.

Over the years Berrebi became involved in more than 300 investments connected to Kima and other vehicles across more than twenty countries. The portfolio from that period includes names such as Wise, Formlabs, Carta, Front, Leetchi and Producteev, alongside companies later acquired by Google, LinkedIn or OpenTable.

The spread did not make him treat companies as a collection of financial tickets.

Berrebi approached investing as compressed co-founding. A founder arrives with twenty problems. The useful work is to find the one blocking the other nineteen, introduce the right person, unlock a hire, reframe a product decision or simply prevent a three-month detour.

That method also came from a personal constraint.

His time was already divided between work, study and an enormous family. He could not spend three days on each company or cross the Atlantic to turn a coffee into an expense line. Scarcity of time became a work architecture.

Kima was therefore more than an unusually active fund. It was a machine built to compress decisions.

He leaves the profession just as he could settle into it

Berrebi left Kima Ventures in 2015 after five years.

It was a strange move. The portfolio was beginning to reveal what it contained. His name had become closely associated with European early-stage investing. Staying would have been the simplest career path.

But Berrebi never really wanted to become only an investor.

He later worked with LetterOne Technology, where the tickets were dramatically larger, and founded Magical Capital. Moving into bigger capital did not resolve the tension. The large-fund world mostly clarified what interested him less: money as the main object rather than as material for building something.

He then stepped away from much of the public noise.

For several years, Torah and Talmud study took almost all of his time. He devoted seven years to full-time study before returning to a split schedule.

This is not a spiritual intermission pasted beside a technology career. The two activities eventually merged in the way he reasons.

Talmud rarely offers a simple answer and moves on. It confronts interpretations, tests edge cases and forces several hypotheses to remain alive before a decision. Berrebi carries that habit into advising and investing: look for the contradiction, restate the problem, attack the assumption that feels obvious.

For many people, spirituality provides an exit from work. In his case, it seems to have changed the way he enters it.

In 2026, the loop returns to product

Today his personal site still presents him as an entrepreneur and investor, with more than 340 investments across 26 countries. Magical Capital remains his investment and advisory platform. A new global investment structure is being prepared.

But the more interesting detail is elsewhere.

Berrebi is building products again.

Coding agents have recreated a situation he knows well: one person, one idea and almost no team at the beginning. He spends hours iterating with models, testing interfaces, changing databases, checking backups, revisiting architecture and starting another pass.

He has even rebuilt, for his own use, a modern version of the Net2one idea: a monitoring system that follows the subjects he cares about and assembles a personal press review. What required a team, expensive servers and custom infrastructure in the late 1990s can now be prototyped by one person in a matter of weeks.

The symmetry is almost suspiciously clean.

Thirty years ago, he was a young journalist fascinated by machines that could filter information. Then he built one. He later invested in people building new machines. He stepped away for years to study a very different body of text. Now he is back in front of a machine, building another system because the one he wanted did not exist in quite the right form.

This is not a straight career

It would be easy to tell Jérémie Berrebi's story as a sequence of numbers: hundreds of startups, dozens of countries, billion-dollar portfolio companies, funds and exits.

That may be the least interesting version of the portrait.

His career looks more like a repeated struggle against work hardening into status.

When Net2one became too large, he reduced it. When Zlio turned a new idea into a machine dependent on Google, he eventually moved on. When investing became an established profession, he returned to study. When study could have become a permanent retreat, he started building again.

Even his relationship with AI follows the same pattern. Models interest him less as magical replacements for labor than as a way to put one person back in direct contact with a product. They make the small team, sometimes the single person, capable of testing an idea before the organization needed to support it already exists.

That may be the most stable line in the whole story.

Jérémie Berrebi has not spent thirty years climbing a ladder.

He has spent thirty years returning to the moment just before someone builds one.