The journey of Gavriel Cohen, creator of NanoClaw, has been nothing short of extraordinary in recent weeks.
Approximately six weeks ago, Cohen unveiled NanoClaw on Hacker News. He positioned it as a compact, open-source, and secure alternative to the widely discussed AI agent-building platform, OpenClaw, a project he conceived and developed over a single weekend coding marathon. This initial post rapidly gained viral traction.
Recounting the intense development period to TechCrunch, Cohen shared, “I sat down on the couch in my sweatpants, and just basically melted into [it] the whole weekend, probably almost 48 hours straight.”
Roughly three weeks later, NanoClaw received another significant boost when a post praising the project from renowned AI researcher Andrej Karpathy also went viral on X.
Just a week ago, Cohen made the pivotal decision to dissolve his AI marketing startup, dedicating his full attention to NanoClaw and establishing a new company, NanoCo, around it. The combined exposure from Hacker News and Karpathy's endorsement had propelled NanoClaw to impressive metrics: 22,000 stars on GitHub, 4,600 forks (indicating new versions being built from the project), and contributions from over 50 developers. Cohen himself has already implemented hundreds of updates, with many more in the pipeline.
Most recently, this past Friday, Cohen announced a strategic partnership with Docker. Docker, a company credited with pioneering the container technology that underpins NanoClaw, boasts millions of developers and nearly 80,000 enterprise clients. The collaboration will see the integration of Docker Sandboxes into NanoClaw.
The genesis of this whirlwind began a few months prior, when Cohen and his brother, Lazer Cohen, launched an AI marketing startup. This venture delivered marketing services such as market research, go-to-market analysis, and blog content, leveraging a small team augmented by AI agents.
The agency quickly secured clients and, according to the brothers speaking with TechCrunch, was on track to achieve $1 million in annual recurring revenue.
Cohen, a seasoned computer programmer with previous experience at website hosting giant Wix, remarked, “It was going really well, great traction. I’m a huge believer in that business model of AI-native service companies that have margins and operate like a software company but are actually providing services.”
He had personally constructed the startup's AI agents, primarily using Claude Code, each engineered for specific tasks. However, a critical "piece" remained elusive. While agents could perform work when prompted, human users lacked the ability to pre-schedule tasks or integrate agents with team communication platforms like WhatsApp for assignment (WhatsApp serving a global function akin to Slack in corporate America).
Cohen then discovered OpenClaw, a popular AI agent tool whose creator is now affiliated with OpenAI. He adopted OpenClaw to develop the missing interfaces and was initially highly impressed.
He described a significant "aha moment," realizing, “this is the piece that connects all of these separate workflows that I’ve been building.” This epiphany led to an immediate desire for more agents: “on R&D, on product, on client management,” envisioning one for every operational task within the startup.
However, this initial enthusiasm soon gave way to profound concern regarding OpenClaw.
While investigating a performance anomaly, Cohen inadvertently uncovered a file where the OpenClaw agent had downloaded all his WhatsApp messages, storing them in plain, unencrypted text on his computer. This included not only work-related communications for which it had explicit access but also his entire personal message history.
OpenClaw has been widely criticized as a "security nightmare" due to its method of accessing memory and account permissions, making it notoriously difficult to restrict its data access on a machine once installed.
While the security concerns of OpenClaw might diminish with time given its popularity, Cohen harbored another significant worry: its immense code footprint. During his investigation into security options, he observed the vast array of packages bundled within it. This included an "obscure" open-source project he himself had developed months earlier for editing PDFs using a Google image editing model, a project he hadn't actively maintained and was unaware was integrated.
He ultimately concluded it was impossible to thoroughly validate OpenClaw's entire codebase and its dependencies, which, by some estimates, extended to over 800,000 lines of code.
In response, Cohen engineered his own solution, NanoClaw, comprising a mere 500 lines of code. Initially designed for internal use within his company, he subsequently shared it publicly. Its architecture was based on Apple’s new container technology, which establishes isolated environments, preventing software from accessing any machine data beyond what it is explicitly authorized to use.
A couple of weeks after NanoClaw's debut on Hacker News, Cohen's phone began ringing incessantly at 4 a.m. A friend, having seen Karpathy’s post, urged Cohen to engage on Twitter, which he did, sparking a public dialogue with the esteemed AI researcher.
The attention on NanoClaw subsequently surged. This included numerous tweets, positive YouTube reviews from fellow programmers, and various news features. The project's burgeoning popularity even led to a domain squatter registering a NanoClaw website URL, though the official site remains nanoclaw.dev.
Following this surge, Oleg Selajev, a developer at Docker, reached out. Recognizing the buzz, Selajev modified NanoClaw to incorporate Docker’s competing container solution, Sandboxes, in place of Apple’s technology.
Cohen readily embraced the integration of Docker Sandboxes into the main NanoClaw project. He reflected on his decision, thinking, “This is no longer my own personal agent that I’m running on my Mac Mini. This now has a community around it. There are thousands of people using it. Yeah, I said, I’m going to move over to the standard.”
Despite the rapid transformations for Gavriel Cohen and his brother Lazer, who now serve as CEO and President of NanoCo respectively, one crucial aspect remains to be defined: NanoCo's revenue model.
NanoClaw is, and the Cohens pledge it will always remain, free and open source. They acknowledge that deviating from this commitment would be seen as a betrayal by the open-source community. Currently, the Cohens are sustaining operations through a friends-and-family fundraising round.
While they are proceeding cautiously with announcing their commercial strategies, largely because these plans are still in development, they report that venture capitalists are already expressing keen interest.
The envisioned commercial strategy involves building a fully supported product offering enterprise services, including what are termed “forward deployed engineers.” These specialists would embed directly with client companies to assist in building and managing their systems, likely focusing on the secure deployment and maintenance of AI agents. However, this is a rapidly expanding and increasingly competitive market.
Nevertheless, given the expansive developer community NanoClaw has now engaged through its Docker partnership, further developments are anticipated in the near future.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.