François-Xavier Leduc knows what two months can cost

On February 14, 2017, François-Xavier Leduc learned that an investor in his car-sharing startup would not have the promised money until mid-April at the earliest. The financing had been expected to close on February 24. For a business already balancing on its remaining cash, two months was a very long time.

Tripndrive entered judicial liquidation on March 16. Leduc explained the ending publicly, setting out the arithmetic of cash in the bank and money spent each month. The idea had customers. The business had locations. Neither could make an investor’s calendar move faster.

Nine years later, Leduc was discussing another financing round, this time for DeepIP, the patent AI company he leads from New York. In March 2026, it announced $25 million in Series B funding, bringing its disclosed total to $40 million. A substantial portion of the previous round remained available, he said. He wanted the financial capacity to accelerate for the next 18 to 36 months.

It would be convenient to draw a straight line between those two moments. Leduc has not presented them as a tidy lesson with a ribbon around it. They do, however, belong in the same portrait: an entrepreneur who has experienced the difference between having an opportunity and having enough time to pursue it.

Before the patent, there was the parking space

The original Tripndrive proposition had the appeal of a well-spotted absurdity. Some travelers paid to leave cars at airports while other travelers paid to rent cars nearby. A vehicle could spend its holiday doing rather more than its owner. Leduc and his co-founders, Arthur de Keyzer and Nicolas Cosme, proposed joining those two journeys.

Tripndrive began operating at Orly in October 2013, followed by Charles de Gaulle in February 2014 and Bordeaux that April. Owners received free parking in return for making their cars available to other travelers. The company handled the rental and insurance. Convenience mattered as much as the economics: customers did not have to negotiate a handover with a stranger carrying a suitcase.

Leduc came to this with a commercial background. His earlier work included Procter & Gamble and Saint-Gobain. His education had included contemporary history at Paris I Panthéon-Sorbonne and business studies at emlyon. On his professional profiles, the emphasis was on marketing, strategy, product and execution. The route into artificial intelligence would come through building businesses.

Tripndrive eventually served about 50,000 users across more than 30 stations and airports. Yet its first model did not grow quickly enough, and the company changed course toward a service for city residents. A useful proposition and a repeatable supply of cars turned out to be different problems. By spring 2017, the cash problem had overtaken both.

TravelCar announced its acquisition of Tripndrive in April 2017, integrating eight employees and the user base. The sequence matters: liquidation came first. Describing the venture simply as an acquisition would remove the part that explains the stakes. Afterward, Leduc worked as an independent consultant for Boston Consulting Group, bringing experience that had become rather more expensive than a business-school case study.

The history student and the data scientist

In 2018, Leduc co-founded Kili Technology with Edouard d’Archimbaud. His partner had led the data science and AI laboratory at BNP Paribas CIB. Their starting point was a practical frustration from applied AI: preparing the data could consume an enormous share of the work before a model produced anything useful.

Kili helped companies turn raw material into annotated data suitable for machine learning. Contracts, images, text and recordings needed organization and human judgment. The promise of an algorithm depended on the quality of what went into it. There was plenty of attention on the engine; Kili worked on the fuel.

Leduc’s contribution was to make that proposition into a business that enterprises would buy. In July 2021, Kili announced a $25 million Series A led by Balderton Capital, alongside existing investors Serena and Headline. It followed a seed round earlier that year. The work of labeling data had acquired a substantial budget.

The Atlantic was already part of the plan. By early 2022, Kili had opened a New York office and recruited leadership for its American expansion. Leduc was pursuing a large US market while the company retained its French base. DeepIP would later use a similar arrangement, with its CEO in New York and its technical co-founder in Paris.

The useful continuity here is specific. Kili dealt with the preparation that makes AI usable; DeepIP deals with the specialist work in which AI must prove useful. The two businesses ask different things of software, but both put Leduc between technical capability and a customer’s working day.

A new technology walks into Microsoft Word

By January 2024, Kili was launching an AI patent copilot called davinci. DeepIP emerged from that work, with Leduc and d’Archimbaud again serving as CEO and CTO. Generative AI offered a way to assist with the dense text that passes between inventors, patent professionals and patent offices.

Patent drafting was a particularly revealing test. The task involves explaining an invention precisely and describing the protection being sought. An impressive paragraph is a small part of the job. A practitioner has to understand the technology, keep terminology consistent and decide what the application should cover. Software has to earn its place beside that judgment.

Leduc’s product decision was to enter the environment where practitioners were already working, including Microsoft Word. Asking a busy professional to abandon a familiar document to visit another application adds a small toll to every use. Put the assistance beside the document and the proposition becomes easier to try. Even technological change occasionally benefits from good manners.

In his account of DeepIP’s early days, Leduc recalled warnings that attorneys would distrust new tools and that the industry would move slowly. He described concentrating on the people doing the work and waiting for the market to become receptive. The patience in that account is paired with a very ordinary commercial discipline: find out what customers will actually use.

The public conversations around the product have included Barry Schindler of Greenberg Traurig and André Marais of Schwegman, Lundberg & Woessner. These are practitioners discussing the work they know, alongside an entrepreneur selling software for it. The relationship gives Leduc a continuing test of his proposition: a feature must survive contact with an actual matter.

What happens to an hour you get back?

A familiar sales pitch for AI begins and ends with speed. Leduc’s more interesting argument concerns what a professional does with the time saved. In discussions of DeepIP, he has described attorneys putting some of that time back into the quality of the finished application. Drafting faster can create room to think longer.

That distinction changes the appeal of the product. A firm working within a fixed fee has limited time to spend on a matter. If software takes care of repetitive drafting and consistency work, more of that budget can go toward the decisions that require experience. The economic attraction is clear; better outcomes remain something practitioners must assess in their own work.

By March 2026, Leduc was emphasizing the accumulation of context across a patent’s life. “But patent work is cumulative,” he said. A draft becomes an application, an application draws questions, and decisions made early continue to matter later. A collection of fast tools can still leave someone carrying information from one place to another.

“But patent work is cumulative.”

François-Xavier Leduc · March 2026

For Leduc, the ambition extends across those stages. DeepIP reported serving more than 400 law firms and corporate IP teams in 25 jurisdictions at its March financing. Those figures describe company-reported adoption at that moment. His task as CEO is to turn that spread of customers into a coherent product, while each team brings its own documents, habits and expectations.

The next stop is inside a German law firm

In June 2026, DeepIP announced the acquisition of PatentMaker. The German business had grown within the law firm Boehmert & Boehmert. Its founder, Matthias Hofmann, joined DeepIP with responsibility for Germany, while managing director Celia Wei led the German operation. DeepIP established its Munich office within the firm’s premises.

The location is an unusually literal expression of Leduc’s product strategy. A company promising to fit into practitioners’ working lives now had an office inside a patent firm. Global expansion had taken it closer to a particular set of desks.

The acquisition also exposed the limits of treating patent work as one uniform market. Local practice and jurisdiction matter to the people using the software. Buying PatentMaker brought DeepIP people who understood those routines. For Leduc, expansion involves learning the differences as well as adding customers.

His public profile remains that of a working founder: product conversations, practitioners, investors and the next market. On Wellfound, he described himself as pragmatic and focused on execution. That self-description fits the subjects he keeps returning to in interviews, where the mechanics of adoption tend to occupy more space than predictions about distant machines.

The parking spaces, the training data and the patent documents make an unlikely set. Each contains work or value that is easy to overlook while looking at the headline attraction: the trip, the model, the invention. Leduc has repeatedly chosen to build around what happens in between.

In 2017, his company could not wait for the money to arrive. In 2026, he was raising money before the earlier round had been spent, while selling a product intended to give professionals more useful hours. The parallel needs no claim about what he privately learned. His career supplies the dates. Time, in Leduc’s businesses, has always had a price.