London-based Trent AI, an agentic safety firm, in the present day introduced its €11 million ($13 million) Seed spherical because it emerges from stealth with a layered safety answer constructed for the agentic period.
The spherical was led by LocalGlobe and Cambridge Innovation Capital, with participation from OpenAI Member of Technical Workers Joaquin Quiñonero Candela, former Head of Stripe Information Infrastructure and present Director at AWS Avinash Bhat, Databricks Distinguished Engineer Ippokratis Pandis, and former Spotify VP Engineering and Head of AI/ML Tony Jebara.
“Organisations are deploying AI brokers and autonomous workflows sooner than their safety can adapt, and most growth groups utilizing these brokers and workflows haven’t any safety framework designed for his or her methods,” says Eno Thereska, co-founder and CEO of Trent AI.
“This isn’t a simple drawback to unravel. Trent AI is tackling these tough and essential issues, whereas constructing the mandatory safety foundations and frameworks for agentic methods now and thru the following decade.”
Within the closest EU-Startups protection from 2025–2026, different related corporations embody:
- Overmind, a London-based startup that raised €2.3 million to develop technical groups, speed up product growth and scale go-to-market for its AI-agent supervision layer
- Innerworks, additionally in London, which raised €3.7 million to increase its AI-powered fraud detection platform
- Cyb3r Operations, one other London firm, which raised €4.6 million to sort out third-party cyber threat
- Qevlar AI, based mostly in Paris, which raised €25.8 million to automate Safety Operations Centre investigations with agentic AI
- Zepo Intelligence, which raised €12.8 million to guard workplaces from AI-driven human-targeted cyber threats
- Galtea, from Barcelona, which raised €2.7 million for its AI analysis platform targeted on dependable AI brokers in manufacturing.
Along with Trent AI’s spherical, that brings sector funding to nearly €63 million, suggesting a visual 2025–2026 stream of capital into European corporations constructing safety, supervision and reliability tooling round AI methods. The UK stands out specifically, as three of the closest comparable rounds are additionally London-based, inserting Trent AI in a notably energetic home cluster.
“Agent adoption is outpacing enterprise safety readiness. As autonomous workflows make choices throughout essential methods, a brand new layer of infrastructure is required to control, observe, and implement secure habits,” provides Cambridge Innovation Capital Associate Ian Lane. “We imagine Trent AI is nicely positioned to outline this class.”
Based in 2025, Trent AI goals to redefine agentic AI with context-driven agentic safety. Its proprietary judgement layer and reinforcement studying know-how energy a set of specialised safety brokers. By orchestrating these brokers throughout buyer workflows, Trent AI transforms agentic safety right into a steady a part of agent growth.
Trent AI was launched by co-founders: Eno Thereska, former Distinguished Engineer at Alcion (acquired by Veeam), AWS, and Confluent, Neil Lawrence, DeepMind Professor of Machine Studying on the College of Cambridge and former Director of ML at Amazon, and Zhenwen Dai, former Machine Studying Scientist at AWS and Senior Analysis Supervisor at Spotify.
“The rise of brokers goes hand in hand with the rise of recent safety threats,” says Saul Klein, co-founder and Government Chairman of Phoenix Court docket, the house of LocalGlobe. “Now’s the suitable time to construct the long-term foundations of safety for agentic methods. Trent AI is uniquely positioned to do that, combining deep tutorial experience with real-world expertise constructing large-scale methods and dealing intently with design companions deploying brokers in the present day.”
In response to Deloitte’s 2026 State of AI report, almost 3 in 4 (74%) corporations plan to deploy agentic AI inside two years. Regardless of this, just one in 5 (21%) report having a mature mannequin for governance of autonomous brokers. The risk will increase in advanced environments with interconnected brokers, with safety holes risking the whole infrastructure; driving the necessity for a safety and compliance answer that encompasses the whole agentic ecosystem.
Trent AI’s layered, unified providing reportedly secures brokers all through the whole lifecycle. Each cycle makes Trent AI’s brokers smarter concerning the methods they defend. Because the suggestions loop tightens, judgement improves and mitigations turn out to be extra correct, giving growth and safety groups a sooner, extra dependable path to secure deployment.
In response to the corporate, brokers within the loop work to:
- Scan: These risk scanning brokers observe code, infrastructure, dependencies, brokers and runtime habits, studying the place threat lives within the ecosystem, laying the framework for safety by design.
- Choose: These evaluation brokers decide and classify sign vs. noise, assess enterprise affect and prioritise based mostly on actual threat moderately than static guidelines. This judgement turns into extra predictive over time.
- Mitigate: These remediation brokers patch vulnerabilities, open pull requests, alter configurations and validate that fixes work for a more healthy code base.
- Consider: These safety posture brokers monitor traits, quantify threat over time, benchmark in opposition to requirements and determine systemic weaknesses for a compounding framework that turns into more and more extra correct at conducting threat forecasts.
Corporations with early entry to the Trent AI agentic answer, together with Cover, Commscentre, ML@Cam, Qbeast, and Weblogic , have reported visibility into their safety posture, a safety audit report, quick response time figuring out and presenting vulnerabilities, a clear and nicely laid out remediation scope and adaptive suggestions.
“Agentic methods are rapidly changing into a part of the software program stack, however the safety infrastructure round them continues to be early,” says Avinash Bhat, Director at AWS. “Trent is constructing the foundations groups will want, to function these methods safely at scale. I’m excited to assist Eno and the workforce as they sort out this rising problem.”
