NanoCo founder Gavriel Cohen officially rejected a $20 million buyout offer for his AI agent framework NanoClaw and instead closed a $12 million seed round led by Valley Capital. The funding specifically targets the 2026 enterprise rush for secure, one-to-one professional assistants that can operate inside corporate networks without leaking data or executing unauthorized commands.
Corporate IT departments currently face an absolute deadlock. They desperately want to deploy AI “second brains” to automate tasks in Slack and Microsoft Teams, but they refuse to install systems vulnerable to malicious prompt injections. NanoClaw bypasses this gridlock entirely by shifting security away from the AI models and down to the actual infrastructure layer.
Rebuilding the AI Sandbox
NanoClaw originally started as a viral 48-hour side project. Cohen built it as a hyper-minimal alternative to the standard OpenClaw framework, which is bloated with over 400,000 lines of code. By contrast, NanoCo stripped its core logic down to roughly 500 lines of TypeScript. A human security team can audit the entire system in about eight minutes.
Every agent runs strictly inside isolated MicroVM-based Docker Sandboxes, utilizing an infrastructure partnership established in March 2026. The actual security mechanism relies on a OneCLI Rust Gateway. This gateway intercepts sensitive outbound API requests, like modifying cloud infrastructure or deleting emails, and freezes them. It then sends an interactive card to a human user in Slack or Teams, requiring a physical “Approve” click before the AI can proceed.
The Race for Enterprise Deployment
The market for security-first AI agent harnesses is highly contested. NanoClaw directly competes with rival frameworks like ZeptoClaw, Lobu, and NEAR AI’s IronClaw. All of these companies are racing to solve the exact same engineering hurdle: moving autonomous AI agents from experimental toys into fully trusted enterprise workers that cannot accidentally destroy a company’s database.
NanoCo is currently booking enterprise customers to deploy its managed assistant platform across these corporate communication tools.