Investor Trust
Demonstrate a mature security posture
Securing France's First Autonomous AI Agent Platform
How RedSentinel helped a Lyon-based deeptech startup validate their security infrastructure before large-scale deployment.
- 7 daysAudit duration
- 150+Tests performed
DigitalKin
Pioneer of Agentic AI in France
DigitalKin is a Lyon-based startup founded in 2023 by Emmanuel Théry (ex-Michelin, ex-Dynergie), Sébastien Deschaux and Nicolas Lester Reynolds. The company develops France's first platform for recruiting, training and managing autonomous AI employees: the "Kins".
Unlike traditional chatbots, Kins are cognitive agents capable of executing complex end-to-end tasks: scientific literature reviews, document analysis, structured report writing, business process automation.
Key Figures
- €400,000 raised in September 2024
- Headquarters : Lyon (69009)
- Reference clients : Laboratoires Boiron
- Awards : French Tech Seed, NVIDIA
- Technology : KinOS / KinConnect
"Our Kins don't replace human expertise, they multiply it."
The Context
Why a security audit was essential
B2B Requirements
Companies like Boiron require security guarantees
Sensitive Data
Kins handle strategic documents (R&D, CIR)
Innovative Architecture
Multi-agent platform integrating multiple LLMs
Commercial Expansion
Slush Helsinki and European scaling
Rare Expertise
Web + AI/LLM dual competence needed
Our Mission
A two-part complementary audit
Web Application Audit
Based on OWASP Top 10, covering authentication, access controls, server configuration, injections and data exposure.
OWASP Top 10
AI/LLM Security Audit
Based on OWASP LLM Top 10, analyzing AI agent-specific risks: prompt injection, data leakage via model, unexpected agent behaviors.
OWASP LLM Top 10
Our Approach
Structured methodology
Preparation
Scope definition, access setup, application mapping
Reconnaissance
Passive infrastructure analysis and entry point identification
Active Tests
Controlled exploitation of potential vulnerabilities
Validation
Finding confirmation and real impact assessment
Delivery
Detailed report and presentation session
Results
What the audit revealed
Vulnerabilities by Severity
- Critical0
- High0
- Medium2
- Low / Info5
Security Score
75/100
Good security posture
Conclusion
No critical vulnerability allowing immediate system compromise. The platform rests on a solid technical foundation. Recommendations mainly focus on strengthening existing configurations, demonstrating already well-established security maturity for a scaling startup.
All vulnerabilities were fixed by the DigitalKin team
Deliverables
A modern audit experience
Real-time Platform
Audit progress visible by client, findings documented on the fly
Executive Report
Summary for management and investors
Technical Report
Each finding with proofs, CVSS scoring, remediation
Priority Matrix
Ranking by criticality and remediation effort
Presentation Session
Live presentation to technical team
Electronic Signature
Signed report via our platform for evidential value
Client Testimonial
"We called on RedSentinel and Alexandre Tavares for a security audit of our platform hub.digitalkin.ai. The collaboration was smooth and very professional: the audit methodology is clear and reassuring, and we were able to follow progress step by step directly from their platform. Alexandre was available, educational and proactive on corrective measures to implement. We leave with a very structured report and a concrete roadmap to strengthen our security. I highly recommend RedSentinel to any tech team that wants to take their product security seriously."
See on LinkedIn
Why DigitalKin Chose RedSentinel
Web + AI/LLM Dual Expertise
A rare expertise combining OWASP Top 10 (web) and OWASP LLM Top 10 (AI) methodologies.
Offensive Approach
We think like an attacker to protect like a defender.
Total Transparency
Real-time audit tracking, no black box.
Actionable Deliverables
Contextualized recommendations, not generic 200-page reports.
Ready to Secure Your Application?
Developing a SaaS application, an AI platform, or a digital product? Identify your vulnerabilities before an attacker does.