Enterprise security spent two decades learning to ask who a person is. The question that now keeps chief information security officers up is what an autonomous agent is allowed to do next.
Cygrid emerged from stealth with a €2.66M pre-seed round led by Atlantic, the pan-European pre-seed fund, with D11Z and angels including Zscaler product vice president Boris Gorin, bonify chief executive Andreas Bermig and AmberVale Capital’s Vlad Tropko.
One graph for humans and agents
Founded in January 2026 by Dmitry Kharchenko and Dekel Paz, both veterans of Israeli military intelligence, Cygrid maps human, machine and AI identities into a single intelligence graph. The platform monitors agent behaviour continuously, tests each action against the permissions its identity holds and blocks or narrows anything that breaches policy before execution.
That is a narrower promise than the dashboard-driven category of agent observability tools. Cygrid is selling an enforcement point, not a viewing window, which puts it closer to identity infrastructure than to monitoring software.
Early traction and the road ahead
The company says it is working with 19 design partners in financial technology, ecommerce and software, six of which run the platform in production. The funding covers engineering hires, deeper enforcement capability and a European go-to-market push.
Timing helps. Agents now hold credentials and session access across corporate systems, and incident research has found that dormant agents keep their connections long after anyone stops using them. Securing the credential is table stakes; deciding which actions it may take is the harder problem, and the pre-seed bar for solving it is still low enough for a two-founder team to clear.
