Trent AI, founded by former AWS engineers Eno Thereska and Neil [surname cut off in source], raised $13 million in seed funding led by LocalGlobe and Cambridge Innovation Capital. The London-based startup launched yesterday with backing from executives at Databricks, Stripe, and other tech companies, positioning itself as an "AI agent security" company.
The timing makes sense—AI agents are moving from demos to production, handling everything from customer service to code generation. But the security landscape for these systems is still being defined. Traditional cybersecurity focused on protecting data and networks. AI agent security involves new attack vectors: prompt injection, model poisoning, data extraction through clever queries, and agents acting outside their intended boundaries. The question is whether Trent AI has identified specific, valuable problems or is betting on a category that doesn't quite exist yet.
The sparse details in the announcement are telling. No mention of specific products, target customers, or even which types of AI agents they're securing. For a security company, that's either strategic stealth mode or they're still figuring out what they're building. The AWS pedigree suggests they understand infrastructure-scale problems, but AI agent security requires different expertise than traditional cloud security.
Developers deploying AI agents should focus on the basics: input validation, output filtering, and clear agent boundaries. Until we see what Trent AI actually ships, the best security is careful prompt engineering and robust monitoring of what your agents actually do.