Fleuret AI raises €4M Pre-Seed for continuous autonomous penetration testing

Fleuret AI (Paris) has raised €4 million in a Pre-Seed round led by RAISE Ventures, with Auriga Cyber Ventures, Wind Capital, and Better Angle participating. Founded in 2026 by Yanis Grigy (CEO) and Augustin Ponsin (CTO), the company builds an autonomous penetration testing platform that continuously simulates attacker behaviour.

The annual pentest problem

Penetration testing — the practice of hiring specialists to attack your own systems before adversaries do — is one of the most widely adopted security practices in enterprise and regulated environments. It is also, in its traditional form, fundamentally broken as a security signal. A manual pentest produces a point-in-time assessment of a specific environment: the infrastructure that existed during a two-to-four-week engagement window, tested by consultants with the techniques and tooling available at that moment.

Modern environments don't hold still. Cloud-native infrastructure deploys new services daily. Configuration drift is continuous. A patch delayed by two weeks creates and closes a window that no annual test will see. The result is a security programme built on a document that is outdated before it reaches the CISO's desk.

Continuous simulation as the alternative

Fleuret AI replaces the snapshot with a continuous loop. Its AI agents run the full attack chain autonomously — reconnaissance of the attack surface, identification of entry points, exploitation attempts, lateral movement across the network — and surface findings in real time as a live vulnerability feed rather than a periodic report. The approach mirrors what Pentera has built in the US market, but Fleuret's European positioning is intentional: NIS2 compliance requirements are pushing EU organisations toward continuous security assurance frameworks, creating demand for exactly this type of tool.

"Manual penetration testing is a snapshot," said Grigy. "Fleuret AI is a continuous adversarial simulation — the only way to keep pace with real attackers."

What Pre-Seed funds

The €4M goes toward building the autonomous engine — the core AI capability that must reliably replicate creative attacker behaviour on novel environments, not just replay known attack paths. This is the hard technical problem: automated tools that follow scripted playbooks are well-established (vulnerability scanners, DAST tools); autonomous agents that reason about novel configurations and chain together non-obvious exploitation steps are not. That is the gap Fleuret AI is attempting to close.

The week also brought Netsec's €8.9M raise for managed security at SMEs — two French cybersecurity companies raising in the same window, each addressing a different layer of the same structural security deficit.