Trent AI

Trent AI is a London based agentic security company founded in 2025 to secure the AI agents and autonomous workflows that enterprises are now deploying faster than their defenses can adapt. The founding team is unusually research heavy: chief executive Eno Thereska was a distinguished engineer at Alcion, which Veeam acquired, and earlier at AWS and Confluent; chief scientist Neil Lawrence is the DeepMind Professor of Machine Learning at the University of Cambridge and a former director of machine learning at Amazon; and chief technology officer Zhenwen Dai came from machine learning roles at AWS and Spotify. In April 2026 the company emerged from stealth with a thirteen million dollar seed led by LocalGlobe and Cambridge Innovation Capital, joined by angels from OpenAI, AWS, Databricks, and Spotify. Trent cites a governance gap: most companies plan to deploy agentic AI, but few have a mature model for governing autonomous agents.

Trent secures agents throughout their lifecycle with a multi agent architecture. Scanning agents continuously observe code, infrastructure, dependencies, agents, and runtime behavior to learn where risk lives. Analysis agents judge signal from noise, assess business impact, and prioritize by real risk rather than static rules, a judgment that a proprietary reinforcement learning layer sharpens over time. Remediation agents patch vulnerabilities, open pull requests, and adjust configurations, then validate that the fixes hold. Security posture agents track trends, quantify risk over time, and benchmark against standards. The platform is built to catch agent specific failure modes such as prompt injection, tool misuse, unintended actions, data exfiltration, and privilege escalation, and it works at the design and architecture level rather than relying on point in time code scanning.

Trent is a partner member of the OWASP project for large language model application security and a startup partner in Carnegie Mellon's CyLab network, and it lists early design partners including Canopy, Commscentre, and Qbeast. For an engineering or security team shipping its own agents that wants continuous, context driven security embedded in the build process rather than bolted on afterward, Trent is a promising fit. Because the company is fresh out of stealth and running with design partners, a buyer needing a mature, broadly deployed platform, self hosted or air gapped delivery, or open model choice should treat it as early stage, since those are not yet documented.