Security assessments for AI agents and LLM-powered apps

Barrikade is security ops for the next generation of companies: AI-native teams shipping agents, copilots and LLM features into production.

We run authorised, scoped security assessments of AI-enabled systems, combining our agent-driven testing platform with expert review. We look for the ways AI agents actually go wrong:

• Prompt injection, direct and via retrieved documents, emails or web pages

• Tool/API misuse, over-broad permissions and dangerous cross-service actions

• Data exfiltration and sensitive-data leakage

• Application security issues around the AI system

You get a clear report that separates proven exploits from suspected issues, with evidence, reproduction steps, impact and developer-ready remediation guidance.

Especially useful if you sell to hospitals, health services or enterprises and need security evidence before a procurement review stalls the deal.

Backed by UCC IGNITE, Enterprise Ireland and NVIDIA. Running paid pilots with select startups.

barrikade.ai