Trent AI Ltd., a London-based artificial intelligence security startup, launched today with $13 million in seed funding.

LocalGlobe and Cambridge Innovation Capital led the investment. They were joined by executives at Databricks Inc., Stripe Inc. and other tech companies.

Trent AI was founded last year by former Amazon Web Services Inc. engineers Eno Thereska, Neil Lawrence and Zhenwen Dai. Lawrence, a machine learning professor at Cambridge University, is the company’s chief scientist. Thereska is chief executive, and Dai is chief product officer.

Trent AI has built a platform that can find cybersecurity issues in AI agents and the code they generate. The software itself uses AI agents to carry out the task. According to Trent AI, its technology is better at spotting threats than incumbent cybersecurity products that were originally designed to scan conventional software.

The company’s platform is powered by four groups of agents. The first group searches for exploits in customer-developed agents’ code, the third-party tools they use to automate tasks and the underlying infrastructure. For example, Trent AI could point out if an AI automation workflow has access to a sensitive database that isn’t strictly relevant to its work.

Many cyberattacks chain together multiple vulnerabilities to steal data. For example, a hacker might use a firewall flaw to breach a corporate network and a different vulnerability to log into a database within that network. Trent AI says its platform can spot such complex threats by simulating attack paths.

After the first group of agents finds potential issues, a second group ranks them based on their severity. For example, a vulnerability in a financial application might be ranked higher than an insecure development environment without sensitive data. Trent AI generates a natural language overview of each issue that explains how it could be exploited.

The third set of agents that the company ships with its platform generates remediation suggestions. Trent AI can recommend improvements to an application’s code, changes to configuration settings and other modifications. Developers can apply the suggestions with a few clicks.

Trent AI’s remaining agents collect information on how an application’s security changes over time. Software teams can use the data to refine their vulnerability remediation workflows.

According to Trent AI, its platform also provides several tool-specific features. There’s a module that enables users to install Trent AI in OpenClaw and scan the agent for issues such as overly broad access permissions. The platform can also connect to Lovable, a popular AI development tool, and check the applications that users build for vulnerabilities.

Trent AI says its software has already been adopted by several tech firms. The company will use the proceeds from its seed round to expand its customer base and engineering team. 

Photo: Unsplash