Israeli AI start-up NanoCo brings governance and security to enterprise AI agents

As companies rush to adopt artificial intelligence tools across their organizations, Israeli start-up NanoCo bets the next phase of enterprise AI will require more than powerful models. Founded by brothers Lazer Cohen and Gavriel Cohen, the company focuses on building security, governance, and control layers around AI agents that can carry out tasks on employees' behalf.

Speaking to The Jerusalem Post, Lazar Cohen said that NanoCo emerged from the founders' earlier work in AI-driven communications technology. 

About a year and a half ago, as interest in agentic AI and AI-powered software development accelerated, the two brothers began building what he described as a native AI public relations agency. The venture attracted clients and helped the team raise a pre-seed round for an earlier company called Quibit. During that period, the developers increasingly relied on AI agent platforms to automate recurring work.

Their experience with one of those platforms, OpenClaw, ultimately led to NanoCo's creation.

"We started using it and realized that the question was whether it could really be trusted with company data," Cohen told The Post.

Cohen leads commercial operations, while Gavriel serves as CEO and oversees product R&D.

The two viewed the challenge as both technical and operational. “We saw value in AI agents capable of completing tasks independently, but we were concerned that existing systems lacked sufficient controls for enterprise deployment,” Lazar said.

30,000+ stars on GitHub

NanoCo's answer was NanoClaw, an open-source AI agent platform designed around a simplified architecture and security-focused deployment model. Cohen explained that the platform uses a much smaller codebase than some alternatives and runs agents in isolated sandboxes designed to prevent them from accessing unauthorized data or systems.

“OpenClaw had half a million lines of code, which is impossible to audit compared to NanoCo’s 4000 lines of code,” Cohen explained.

In addition to the sandboxing technology, NanoCo developed a gateway that sits outside the AI agent environment. Requests such as accessing information or sending communications are evaluated through that gateway, where a human operator either approves or denies the request.

Cohen said the company's open-source release gained traction quickly, was adopted by hundreds of thousands of users, and accumulated more than 30,000 stars on GitHub. He added that the adoption helped attract investor attention and supported the company's growth strategy.

In 2026, NanoCo announced a $12 million seed funding round led by Valley Capital Partners, with participation from investors including Docker, Vercel, and monday.com.

In addition to NanoClaw, its open-source platform, the company’s commercial enterprise offering, NanoCo, has also been gaining traction worldwide.

Cohen told the Post that NanoCo was endorsed by Andrej Karpathy (who co-founded OpenAI) and by Singapore's Foreign Minister, Vivian Balakrishnan. Venture capitalists from leading funds have also been using the platform.

Daniel Ziv, an investor and principal at Glilot Capital Partners, explained that “a VC’s scarcest resource is attention, and a lot of our time goes to meta-work rather than judgment: scheduling, writing notes, updating the CRM, chasing diligence items, remembering what a founder said three months ago. Agents absorb that work, and because they’re always on, nothing falls through the cracks between meetings.”

Glilot is an AI-first firm instituted in Israel in 2011 as the world's first venture capital company dedicated to cybersecurity. According to Ziv, the security and privacy provided by NanoClaw have been paramount.

“We were Israel’s first cybersecurity fund, so we couldn’t compromise here. A VC handles extremely sensitive information, and NanoClaw is a security-first, privacy-aware agent infrastructure, which made it something I could actually trust with my data.”

And, with the agents doing the footwork, “that leaves me free for what stays human: creativity, like finding a smart way through a deal conflict or helping a portfolio company out of a tough spot, and relationships with founders, advisers and customers. My back-office job has shifted from doing the tasks to writing the workflows,” Ziv added.

For enterprise organizations

While NanoClaw is designed primarily for individual users and developers, NanoCo packages the technology with additional governance, management, and security controls intended for organizations.

According to Cohen, many companies do not want to manage complex AI infrastructure internally. Instead, they are seeking turnkey solutions that can be deployed and governed within existing corporate environments.

NanoCo's enterprise platform, he explained, is deployed directly into a customer's cloud environment, allowing organizations to maintain ownership and control over their data. The company does not have access to a customer's source code or internal information beyond permissions explicitly granted by the customer. NanoCo can manage enterprise deployments, including software updates and operational support. Customers can also move their agents, stored knowledge, and workflows between providers if they choose to switch AI models or infrastructure platforms.

"Companies sometimes feel like they're being left behind," Cohen said. "But it is still very early."

Tiang Lim Foo, a founding partner at Analog Ventures, said that the company built its internal agent platform, Amplify, on NanoClaw.

"Our AI analysts work in Slack alongside the team, covering investment triage, portfolio management, and investor relations. It lets a lean seed fund move like a much larger firm,” he said, adding that their agents “handle confidential founder and investor data, and NanoClaw is small enough to audit end to end, with each agent sandboxed in its own container."

NanoCo employs approximately 20 people across three countries. Most of the workforce is based in Israel, where it conducts all research and development. Team members in Spain maintain the open-source NanoClaw project, while personnel in the United States focus on marketing and growth initiatives.

The start-up is entering a crowded AI market that includes consumer-facing assistants and enterprise productivity platforms. However, Cohen said many organizations remain in the early stages of adoption.

For NanoCo, the estimation is that demand for AI agents will keep rising as businesses move from experimentation to deploying systems that can complete work autonomously under defined governance and security frameworks.

"The use of AI will only increase," Cohen said. "We are only at the beginning."