Here is the problem NeuralTrust is selling against, and it’s a good one: inside a large company, AI agents are now breeding faster than anyone is tracking them. One team builds a customer-service agent on one model, another automates back-office work on a different one, and the vendor software you already bought quietly ships with agents of its own. They connect to internal systems, they call outside tools, and they act on their own. Ask a security team how many they’re running, and most can’t tell you.
The Barcelona company just raised €17.2 million to fix that, a round it describes, plausibly, as the largest cybersecurity seed ever raised by an EU company. Alstin Capital led, with VentureFriends, Seaya, Kibo Ventures and Banc Sabadell joining, plus public money from the European Innovation Council and Spain’s research agency.
The statistic worth keeping is operational, not financial. NeuralTrust says it inspects millions of agent interactions a day, and that around 1.2% are malicious, attempts to extract data, hijack a tool, or push an agent past its rules. That’s not a hypothetical risk deck. That’s a live attack surface, caught and stopped in real time.
Gartner reckons 40% of enterprises will quietly decommission AI agents by 2027 because of governance gaps found only after something breaks. NeuralTrust’s whole business is being the thing you install before that happens.
Co-founder Joan Vendrell’s framing is the standard-but-correct one: “AI agents are now part of enterprise operations, but the controls protecting them are still catching up.” The product is three parts, a gateway that brokers every model and tool call, a runtime engine that stops threats live, and a posture layer that finds every agent in the building. Customers already include Iberia, Air Europa, Abanca and Banc Sabadell, which is exactly the conservative, regulated buyer base you’d want validating a security pitch.
The bet is that “secure the agents” becomes a budget line as inevitable as “secure the laptops” did. The risk is that the giant platforms, the ones selling you the agents, decide to bundle the safety controls themselves. NeuralTrust’s answer is to be the neutral control layer that works across all of them, which is a defensible place to stand, right up until it isn’t. For now: a Barcelona company doubled its 2025 revenue in a single quarter and just raised Europe’s biggest security seed to babysit everyone else’s robots.