Fleuret AI Raises €4 Million to Automate Penetration Testing

Fleuret AI, a French offensive cybersecurity startup founded this year, has closed a €4 million pre-seed round to automate a large part of penetration testing with AI agents, Le Monde Informatique reported on October 6. RAISE Ventures led the round, with Auriga Cyber Ventures, Wind, Better Angle and several business angels taking part.

Penetration tests, or pentests, reproduce an attacker’s techniques in a controlled setting so that a company can find and fix its weaknesses. They still depend heavily on specialists, who are scarce and expensive in France. Co-founders Yanis Grigy, the chief executive, and Augustin Ponsin, the chief technology officer, previously ran a conventional pentest firm, which they started while Grigy was still in preparatory classes, and concluded that much of the repetitive work could now be automated.

Price and speed are the pitch. A traditional pentest can cost between €12,000 and €20,000 and tie up consultants for several weeks, the publication noted. Fleuret AI lists an entry price of €4,000 and promises results within 48 hours. Its platform relies on AI agents that explore a company’s digital environment, identify vulnerabilities and try to exploit them, including two in-house agents, Émile and Champollion, designed to carry a human pentester’s know-how into the automated process. “What we do is fundamentally 90% AI and 10% expertise and cyber-specific business applications,” Grigy said.

What it means in practice

The money will mostly go to engineering. The company has just over ten employees and plans to get close to twenty, and it wants to grow from pentesting into an all-in-one offensive security platform. It already names Stoïk and Brevo among its customers, has clients in Italy and Norway and is considering the United Kingdom. Grigy argues that European cybersecurity rules, which require pentests at many companies, make buyers fairly similar from one country to the next.

The round adds to a wider French focus on AI for cyber defence. Mistral presented its new Large 4 model partly on its defensive cybersecurity skills, and Inria’s €13.5 million evaluation and safety programme targets the testing of AI systems themselves.

What to watch: how auditors and insurers receive automated pentests, how Fleuret AI keeps agents that attempt real exploits within agreed limits, and whether it opens an office in the United Kingdom.