The cybersecurity landscape is fundamentally shifting as artificial intelligence democratizes advanced attack vectors, putting enterprise-grade offensive capabilities into the hands of virtually anyone.
In response to this escalating threat environment, Paris-based startup Fleuret AI has secured €4 million in a pre-Seed funding round to completely overhaul how organizations approach penetration testing.
Led by RAISE Ventures with strategic backing from industry veterans like the founders of Stoïk, GitGuardian, and Vade, the capital injection is earmarked for aggressive talent acquisition across software engineering and offensive security.
Founded in 2026 by Yanis Grigy and Augustin Ponsin, the company is actively moving away from the traditional, episodic model of vulnerability assessment.
Instead, they are building a continuous, sovereign security platform powered by autonomous AI agents designed to permanently outpace modern threat actors.
Shifting from Static Audits to Continuous Defense
Historically, penetration testing has relied heavily on one-off manual audits that provide a mere snapshot of a company’s security posture at a specific moment in time. The structural flaw in this legacy approach is that enterprise environments are highly dynamic.
A routine code deployment or minor infrastructure tweak just days after a successful audit can inadvertently expose a massive new attack surface, rendering the previous security certification entirely obsolete. The regulatory environment is also forcing a change in corporate strategy.
According to a recent report by EU-Startups, stricter compliance frameworks like the NIS2 directive now require European organizations to continuously demonstrate the real-time robustness of their security architectures.
Fleuret addresses this operational gap by transforming penetration testing from a static checklist into an ongoing defensive layer.
Their architecture rests on five continuous pillars: comprehensively understanding exposed systems, identifying vulnerabilities, proving raw exploitability, actively assisting with remediation workflows, and strictly verifying that the deployed patches actually hold up against subsequent attacks.
The Mechanics of Agentic Exploitation
Two autonomous AI agents, Emile and Champollion, power the platform. They actively assess systems from an attacker’s perspective. Instead of running passive scans, they probe applications, APIs, and cloud infrastructure. They uncover complex structural weaknesses across assets.
When the agents detect a vulnerability, they attempt a safe exploit. The system produces definitive proof of compromise for every issue. This evidence eliminates false positives and cuts operational noise. In addition, every new deployment automatically triggers fresh tests across the attack surface.
Fleuret also accelerates the remediation lifecycle. The platform routes findings directly into existing engineering ticketing systems. It prioritizes threats by exploitability and verifies bug fixes automatically after code commits.
Early customers like Brevo and Yogosha already deploy the system. Fleuret aims to make continuous agentic security accessible to organizations of all sizes.
