NanoClaw creators reject $20M buyout, raise $12M seed round in six weeks after viral launch

From Couch to Term Sheet in Six Weeks

NanoCo AI has closed an oversubscribed $12 million seed round for NanoClaw, the secure open source AI agent that went viral after endorsements from AI researcher Andrej Karpathy and Singapore's foreign minister 1

1

. The funding was led by Valley Capital Partners, with participation from Docker, Vercel, monday.com, Slow Ventures, and angels including Clem Delangue, CEO of Hugging Face 2

2

. Brothers Gavriel and Lazer Cohen turned down a roughly $20 million acquisition offer to build their vision of a secure AI agent platform that could transform the future of work.

When Gavriel Cohen sat down on January 29 and wrote the first line of code for NanoClaw, he wasn't thinking about seed funding 3

3

. The former Wix developer was solving a problem: existing agentic tools were powerful but dangerously insecure. Six weeks later, NanoCo AI had a term sheet, making it the first company in the rapidly growing "claw" space to close institutional funding 3

3

.

Turning Down Millions to Build Enterprise-Grade Agentic AI Assistant

The Cohen brothers received multiple acquisition offers during their rapid ascent. One venture capitalist offered a "six-digit" dollar amount to buy the project for a portfolio company 1

1

. About two weeks later, another offer came in for around $20 million, including jobs to stay and run their company. They declined both. A founder friend had given them crucial advice: open-source projects grow exponentially more valuable as their community grows, with users contributing code and discovering new use cases 1

1

.

Gavriel estimated that about 50 or more founders and tech executives sent direct messages asking to invest 1

1

. The viral momentum started when Andrej Karpathy tweeted his praise, but really snowballed after Singapore's Foreign Minister Vivian Balakrishnan called NanoClaw his "second brain" in a Facebook post 1

1

4

4

.

Security Built Into Infrastructure, Not Prompts

NanoClaw was created as a secure OpenClaw alternative to assist the Cohen brothers with their previous AI marketing startup 1

1

. Unlike OpenClaw, which grew to 400,000 lines of code, NanoClaw's core logic was intentionally minimized to roughly 500 lines of TypeScript, allowing human security teams to audit the entire system in about eight minutes 2

2

.

The technical approach centers on running each AI agent inside a sandbox using Docker Sandboxes with MicroVM isolation 2

2

4

4

. Raw API credentials never reach the agent itself. Instead, outbound requests pass through a secure OneCLI Rust Gateway that enforces company-defined policies 2

2

. When an agent attempts sensitive "write" actions like modifying cloud environments or deleting emails, the gateway intercepts the request and pings the human user via interactive cards on Slack, Teams, or WhatsApp 2

2

.

Enterprise Customers Drive Productivity Enhancement

NanoCo AI has started booking enterprise customers after early adopters—many of them executives at major tech companies—kept getting requests from coworkers for help setting up their own instances 1

1

. The company now offers implementation services through "forward-deployed engineers" to help businesses roll out NanoClaw to employees with ongoing support 1

1

.

Executives at Amazon, Gap, Google, Meta, SentinelOne, and Accenture are using NanoClaw 1

1

4

4

. Gavriel Cohen told reporters that countless early adopters, including multiple CEOs, report NanoClaw makes them two-to-three times more productive once they master using it 4

4

. Since launching in February, NanoClaw has achieved almost 29,000 GitHub stars 4

4

.

Open Source Commitment With Commercial Services

Despite the enterprise push, NanoCo AI maintains its commitment to open-source principles. The core NanoClaw framework remains available under the permissive MIT License, allowing developers and companies to fork, modify, and run the system locally without paying NanoCo AI 2

2

. The monetization strategy focuses on enterprises that lack specialized engineering resources to build, maintain, and scale internal agent platforms 2

2

.

As users forward emails, documents, and call notes to their agent, it systematically builds an "LLM wiki" that creates a dynamic knowledge graph of the user's specific job and projects 2

2

. This persistent memory allows the agent to shift from answering questions to actively executing first drafts that rival human output. Valley Capital Partners' Steve O'Hara noted that AI is shifting from a tool that answers questions to an assistant that gets work done, and "NanoCo has taken the lead" with formal partnerships and security-first architecture 4

4

.