NeuralTrust raises $20M on AI agent security

Joan Vendrell, co-founder and CEO of NeuralTrust, has closed a $20 million seed round to build the control layer that tells enterprises what their AI agents are actually doing.

The round was led by Munich-based Alstin Capital, whose Alstin III fund raised €175 million in 2024 and typically writes tickets between €2 million and €7 million at seed. NeuralTrust received a cheque several multiples above that range. Co-investors include VentureFriends, Seaya, Kibo Ventures, Banc Sabadell, the EA Ventures Plug and Play Fund, and Finaves. The European Innovation Council and Spain's State Research Agency (AEI) contributed public grants alongside the private capital.

The problem NeuralTrust sells against is sprawl. Inside a large enterprise, AI agents multiply without any central record. One team deploys a customer-service agent on one model. Vendor software arrives carrying agents of its own. Each connects to internal systems and acts autonomously. Most security teams cannot say how many are running, what each is permitted to do, or whether one has been manipulated into leaking data.

"If you connect AI to your email system and it sends emails to outside addresses, leaking internal information, that's a disaster." — Joan Vendrell, Co-founder and CEO.

NeuralTrust's answer is a single integrated control plane sold as three products. TrustGate is a gateway through which every model call, tool call, and agent interaction passes. TrustGuard is a runtime engine that detects and blocks attacks as they happen. TrustLens maps every agent in an organisation and monitors its behaviour continuously. The company says it currently inspects millions of agent interactions per day and finds that roughly 1.2 per cent, approximately one in 80, are malicious.

NeuralTrust's direct competitors are Protect AI, now absorbed by Palo Alto Networks following a reported $500 million acquisition that closed in July 2025, Enkrypt AI, and SSI. That Palo Alto deal is the clearest signal available that AI agent security has become a standalone acquisition category. NeuralTrust's structural position is geographic: it is the only European-headquartered platform of this type, 80 per cent of its customer base is European, and 92 per cent of customers post revenues above €862 million. That concentration makes it the default compliance-safe choice as EU AI Act enforcement moves from framework to obligation across the high-risk domains, banking, aviation, and government, where NeuralTrust already operates.

Named customers include Abanca, Banc Sabadell, Iberia, and AirEuropa, alongside global banks, energy companies, and government agencies. All of these were signed before the company raised its first institutional round. NeuralTrust has also published original security research that has been incorporated into the standards of the OWASP AI Security Project, giving it research credibility that vendors building on third-party findings cannot replicate.

The founding team reflects the enterprise positioning. Vendrell previously served as Chief Data Officer at Mango after roles at Amazon and McKinsey. Victor Garcia, CTO, led security architecture in regulated telecoms and banking. Alejandro Domingo, COO, led AI transformation as an Associate Partner at McKinsey before scaling a startup to exit.

"AI agents are now part of enterprise operations, but the controls protecting them are still catching up. This round allows us to keep building the infrastructure layer that makes AI adoption measurable, governable, and safe." — Joan Vendrell, Co-founder and CEO.

The broader argument behind the round is timing. Gartner predicts 40 per cent of enterprises will pull back autonomous agents by 2027 over governance gaps identified only after something breaks. NeuralTrust is betting that enough enterprises will buy before the incident rather than after it, and that European regulated institutions in particular will want a vendor whose data residency and auditability do not require a conversation with a US regulator.

What the $20 million unlocks is the scale to become the default compliance infrastructure for European enterprises deploying AI in regulated environments before the next consolidation wave decides who gets acquired and who does the acquiring.