Berlin Startup Lands $3M to Stop AI Agents From Leaking Your Data

Jan de Vries · 2026-10-23

Cygrid, a Berlin-based cybersecurity startup, just stepped out of stealth with a $3 million pre-Seed round (€2.66 million). The company is tackling a problem most businesses don't even know they have: AI agents accessing confidential data without proper oversight. The round was led by Atlantic, a pan-European pre-Seed fund, with participation from D11Z (the venture arm of Schwarz Group) and angels like Boris Gorin (VP Product at Zscaler and founder of Canonic Security), Andreas Bermig (CEO of bonify), and Vlad Tropko of AmberVale Capital. ### The Problem With Traditional Access Control Here's the thing about conventional identity and access management tools: they check your ID at the door and then look the other way. Once you're in, you can do pretty much anything. That worked fine when humans were the main actors. But now, AI agents are running around inside enterprises, and they don't need a login screen to cause trouble. Cygrid's platform enforces limits continuously. It doesn't just grant access at sign-in; it checks every action an agent takes and blocks or narrows it in real time. The company maps human, non-human, and AI agent identities into a single intelligence graph, correlating what each identity is entitled to, what it actually does, and on whose behalf it acts. That means they can revoke a forgotten token in seconds or stop a risky AI agent command before it executes. > "Access control tells you who can log in. It says nothing about what they do next." – Dmitry Kharchenko, CEO and co-founder of Cygrid ### Traction and Timing Cygrid was founded in January 2026 by Dmitry Kharchenko and Dekel Paz, both veterans of Israeli Military Intelligence. In just nine months, they're already working with 19 design partners across FinTech, e-commerce, and SaaS, with six in production. One early customer, a Berlin-based D2C company, discovered AI agents and third-party app connections its own security team didn't know about—and used Cygrid to fix over 100 risky findings. The timing makes sense. Attacks on trusted identities are on the rise. Palo Alto Networks reports that machine identities now outnumber human ones by 109 to 1, with 79 of those being AI agents. A CrowdStrike report from 2026 found that 82% of cyber detections involve no malware at all—attackers simply exploit trusted identities like leaked tokens or compromised agent providers. And with more than 80% of Fortune 500 companies already deploying AI agents, and nearly a third of employees using connected agents without organizational approval (according to Microsoft's Cyber Pulse report 2026), the need for continuous enforcement is urgent. ### What's Next Cygrid plans to use the fresh capital to expand its enforcement capabilities, grow the team, and scale go-to-market operations across Europe. The goal? To become the European alternative to US and Israeli providers of AI identity security. "This funding allows us to scale that approach and help organizations move from reactive identity security to continuous, real-time enforcement," added Kharchenko. In other words, companies can adopt AI agents at full speed without losing control of what those agents can do.