Cygrid, a Berlin-based identity enforcement platform, emerged from stealth on October 22, 2026, with €2.66 million in Pre-Seed funding. The company had already stopped more than 2,000 risky identity actions at over 20 organizations before its public launch. Backed by Atlantic and supported by d11z and security operators, Cygrid aims to help enterprises secure AI agents, employees, and machine identities without slowing AI adoption.
The Growing Identity Enforcement Challenge
Most cyberattacks now begin with a stolen or misused identity rather than malware, according to Cygrid. CrowdStrike's 2026 Global Threat Report found that 82% of detected attacks used no malware, while the fastest attackers moved from first access to other systems in just 27 seconds. Machine identities, including AI agents, now outnumber human identities by 109 to 1, according to Palo Alto Networks.
A Year of High-Profile Failures
Several 2026 incidents show how trusted access can be turned against an organization. At Vercel, a hacked AI app gave attackers a token that acted as a skeleton key to internal systems, and at PocketOS an AI coding agent deleted a production database and its backups in nine seconds. A forgotten credential at Klue exposed customer Salesforce and Gong data, while an internal Meta AI agent exposed company and user data without permission.
Market Momentum and Security Warnings
AI agent adoption is accelerating even as security leaders sound alarms. Microsoft's Cyber Pulse report from March 2026 found that 80% of Fortune 500 companies are already deploying AI agents, and 29% of employees have used agents their company never approved. IDC expects 1.3 billion AI agents at work by 2028, while JPMorgan Chase CISO Patrick Opet and Microsoft security chief Charlie Bell have warned that unchecked token access creates direct risks to sensitive resources.
How Cygrid Enforces Identity Decisions
Cygrid connects to existing systems in read-only mode and builds an Identity Intelligence Graph that maps employees, service accounts, API connections, and AI agents. The graph continuously scores risk and decides what each identity may do before a credential is issued, after it exists, and at runtime. Enforcement includes revoking stale grants, rewriting or blocking AI tool calls through the Cygrid MCP Gateway, and correcting policy drift at its source.
Funding, Traction, and Early Validation
The €2.66 million Pre-Seed round was led by Atlantic, with participation from d11z and angels including Andreas Bermig, Boris Gorin, and Vlad Tropko. Since its founding in January 2026, Cygrid has gained more than 20 design partners across fintech, e-commerce, and SaaS, including Barfer's and Horizn Studios. The platform runs in production on Google Workspace, Microsoft 365, GitHub, Slack, and HubSpot, and the company won Germany's federal AI Nation program grant.
Leadership and Investor Perspective
Founders Dmitry Kharchenko and Dekel Paz both served in military intelligence and bring deep security and technology experience. Kharchenko previously held CTO roles at bonify and Friendsurance, while Paz led security teams at Zero Networks, Novartis, and LightCyber. Investor Boris Gorin said Cygrid enforces rather than simply reports, allowing CISOs and CTOs to adopt AI agents while still stopping the one action that should not happen.
Cygrid enters a market where identity risk has become the top concern for 78% of CISOs, according to Team8. By combining deterministic policy and AI-driven behavioral analysis in a single graph, the company drives pre-use, out-of-band, and in-band enforcement from one control plane. With early production deployments and fresh capital, Cygrid aims to become the foundational enforcement layer for identities in the AI era.