Innovation

Sovereign single sign-on solution

Identity management is the primary attack vector in documented breaches. Available single sign-on (SSO) solutions impose either a dependency on a cloud provider or high operational complexity for internal teams.

We are working on an SSO architecture fully deployable on private infrastructure, with no external service calls, and an administration interface usable by a non-specialist team. The underlying protocols are standard (OIDC, SAML). The objective is to eliminate hyperscaler dependency without increasing operational load.

Autonomous alert processing agent for levels 1 and 2

Security alert triage is repetitive at low levels, but a misclassification at level 1 produces cascading effects. The objective is not to remove the analyst, but to present them with already-qualified, enriched and prioritised alerts, with the relevant context elements.

The agent under development integrates several non-negotiable constraints: technical identifiers are pseudonymised before any processing by a language model, no client data transits through external infrastructure, the inference model runs locally on the client's infrastructure, and every qualification decision is logged with its justification.

Level 3 and above alerts are systematically routed to a human analyst. The agent has no capability to act on infrastructure.

Automated and continuous penetration testing

A penetration test produces a point-in-time measurement on a given surface. The surface evolves: new exposures, newly published vulnerabilities in deployed components, configuration changes. A periodic assessment does not reflect the actual continuous state.

We are working on a system for continuous external exposure assessment, distinct from a conventional vulnerability scanner: it simulates attacker action sequences on an authorised scope, taking application context into account, and produces a differential report between two executions. The objective is to detect posture regressions between formal human audits.

HackMe, bug bounty platform

A platform connecting organisations and a community of security researchers, for continuous evaluation of their systems alongside point-in-time tests. Compensation per validated finding, private or public programmes.