AI news · October 7, 2026

Fleuret raises €4 million for an agentic penetration-testing platform

securityagentsdevtoolsfunding

Fleuret is packaging penetration testing as a repeatable agent workflow: discover, prove, fix, and retest. Its free exposure check and €4,000 one-off web-app test make the first commercial step inspectable.

pre-seed funding
€4M
one-off web-app test
€4,000
free-plan repositories
10

Fleuret announced a €4 million pre-seed round on October 5 and described a security platform built around agents for offensive testing. The company says its agents map an organisation's external systems, find and prove attack paths, help fix the issue, and retest the result. It names two agents, Emile and Champollion.

The public product page offers a free exposure check and a free organisation plan for two users and up to ten repositories. It lists static analysis, software composition analysis, secret scanning, dynamic testing, and exposure checks. A one-off web-application penetration test is priced at €4,000, compared with the €15,000 to €30,000 range the company gives for a traditional firm engagement.

The product puts a bounded paid action behind the agent claim. A buyer can start with an exposure check, inspect the evidence, and decide whether a deeper test finds material issues. The hard part is proof quality: a report that cannot be reproduced, prioritised, and closed is not a security outcome.

What you can do with it

Start with a non-production asset and the free exposure check. Require a reproducible proof, affected surface, suggested fix, retest result, and human sign-off for every finding. Compare the agent's useful findings and time to closure with one known-good manual assessment.

Our take

Fleuret has a clearer wedge than a generic security chatbot because the workflow ends in a retest. The €4,000 price is a testable offer, not revenue evidence. The product earns trust only when it finds issues that matter and reduces the time from discovery to verified closure.

Source: Fleuret ↗ — Made With Models writes the brief; the reporting is theirs.