AI Agents Are Rewriting Penetration Testing — This Paris Startup Just Raised $4.3M

Jan de Vries · 2026-10-05

Penetration testing has always felt a bit like getting a health checkup once a year. You feel fine for a while, then everything changes. Fleuret AI, a Paris-based cybersecurity startup, just raised €4 million (about $4.3 million) in pre-Seed funding to fix that problem. They're using AI agents to turn pentesting from a one-off audit into a continuous, always-on security layer. ### The Problem With Annual Pentests Here's the thing: companies change their systems every single day. New deployments, new APIs, new infrastructure. But penetration testing? Still mostly a snapshot. A point-in-time audit that tells you how secure you were last Tuesday. That gap matters more than ever. Under NIS2 regulations, more organizations have to prove their security is robust. Not just once, but continuously. A single pentest report from six months ago doesn't cut it anymore. ### Meet Emile and Champollion Fleuret's platform runs two AI agents with distinctly French names. Emile and Champollion map a company's environment — applications, APIs, infrastructure — then actively try to exploit what they find. Every vulnerability comes with a proof of compromise, so you know it's real, not theoretical. But finding bugs is only half the battle. The platform also helps you fix them: - Prioritization by exploitability (so you tackle what actually matters) - Integration with tools your engineering team already uses - Automatic re-testing after fixes - Reports that verify a vulnerability is truly closed ### Who's Backing This? RAISE Ventures led the round, with participation from Auriga Cyber Ventures, Wind Capital, Better Angle, and a roster of angel investors who know cybersecurity inside out. Names like Jules Veyrat (Stoïk), Eric Fourrier (GitGuardian), and Georges Lotigier (Vade) — these aren't casual bets. They're people who've built and scaled security companies. "AI has changed the nature of cyberattacks by putting powerful tools within everyone's reach, often for free," said Thibaut Schlaeppi, co-head of RAISE Ventures. "Fleuret built its platform to evolve continuously and stay a step ahead of attackers." ### A European Ambition From Day One Fleuret was founded in 2026 by Yanis Grigy (CEO) and Augustin Ponsin (CTO). The company already employs about ten people and counts Brevo, Stoïk, and Yogosha as customers. Their goal isn't just to automate pentesting as it exists today. It's to build the offensive security layer that keeps companies secure all year round — whatever their size. "From the start, the ambition is European: make offensive security accessible to every company, including those that cannot afford regular penetration tests," the company said. That's the real shift here. Continuous, AI-driven pentesting could finally make strong security affordable for mid-sized companies, not just the Fortune 500. And that's a big deal.