The creators of NanoClaw — the hit open supply, enterprise-friendly variant of autonomous AI agent harness OpenClaw — are shifting in direction of commercializing their expertise for enterprises at scale, aiming to offer them with safe AI brokers, and an ever-updating library of office context, for every human worker the enterprise has authorised.
The duo, together with former Wix.com engineer Gavriel Cohen and his brother Lazer Cohen, additionally founding father of tech public relations agency Concrete Media, shared with VentureBeat that their new startup, NanoCo AI, has acquired a $12 million oversubscribed seed spherical was led by Valley Capital Companions.
The spherical contains a roster of strategic backers that reads like an enterprise infrastructure all-star workforce, together with Docker, Vercel, monday.com, Factorial Capital, and Hugging Face CEO and founder Clem Delangue.
Buoyed by the seed spherical, NanoCo AI desires to maneuver past primary automation to supply each enterprise employee a safe "skilled assistant." But they’re nonetheless dedicated to constructing out and sustaining NanoClaw as an MIT Licensed, enterprise-friendly, open supply customary — simply providing specialised business managed companies integration atop it.
The brand new killer use case: an knowledgeable, ever-updating private assistant for every human employee
Gavriel, now CEO of NanoCo AI, sees this customized method as the last word unlock for the trendy employee.
“The killer use case is the the one to 1 we're calling it skilled assistant,” Cohen defined in a latest unique interview with VentureBeat. "When you can provide somebody an agent and make them twice, thrice as efficient, you then in all probability need extra folks as effectively, proper?"
He famous that as customers ahead emails, paperwork, and name notes to the agent, it systematically builds an "LLM wiki" — much like the "LLM Information Base" idea articulated by influential AI researcher Andrej Karpathy — successfully making a dynamic information graph of the consumer's particular job and initiatives.
This persistent reminiscence permits the agent to shift from merely answering inquiries to actively remodeling info and executing first drafts that rival human output.
Cohen emphasised that NanoClaw acts as a large productiveness multiplier relatively than a headcount substitute.
One-to-one safe 'lobster' AI
NanoCo’s core providing is a one-to-one skilled AI assistant designed to shadow staff, draft contracts, evaluate code, and handle accounts instantly inside instruments like Slack and Microsoft Groups.
Quite than a generic chatbot, the assistant learns the worker's position and adapts to their particular working model by unusual dialog.
How does NanoCo forestall this extremely succesful assistant from going rogue? By shifting safety away from fragile immediate engineering and embedding it instantly into the infrastructure.
Not like its predecessor and inspiration, the even standard open supply AI assistant OpenClaw — which grew to an enormous 400,000 strains of code — NanoClaw’s core logic was deliberately minimized to roughly 500 strains of TypeScript. This minimalism ensures all the system might be audited by a human safety workforce in about eight minutes.
Moreover, each NanoClaw agent operates inside a strictly remoted surroundings. Leveraging a strategic partnership with Docker introduced in March, NanoCo AI runs these brokers inside MicroVM-based Docker Sandboxes.
“In NanoClaw, the 'blast radius' of a possible immediate injection is strictly confined to the container and its particular communication channel,” Cohen beforehand defined.
To forestall unauthorized actions, uncooked API credentials by no means attain the agent itself. As an alternative, outbound requests go by a safe OneCLI Rust Gateway that enforces company-defined insurance policies. If an agent makes an attempt a delicate "write" motion—like modifying a cloud surroundings or deleting an electronic mail—the gateway intercepts the request and pings the human consumer through a wealthy interactive card on Slack, Groups, or WhatsApp.
Solely when the consumer explicitly faucets "Approve" does the system inject the credential. It’s the architectural equal of a extremely succesful junior worker drafting an necessary company communication, however being bodily unable to click on "ship" with out the supervisor turning a literal launch key.
Continued dedication to open supply, MIT License
Regardless of its new enterprise push, NanoCo AI is sustaining its dedication to its open-source basis. The core NanoClaw framework stays obtainable below the permissive MIT License, which means unbiased builders and firms can proceed to fork, modify, and run the system domestically.
In plain phrases, the MIT License permits anybody to make use of the software program commercially with out paying NanoCo AI, supplied they embody the unique copyright discover.
NanoCo AI's monetization technique as a substitute focuses on the overwhelming majority of enterprises that lack the specialised engineering sources to construct, preserve, and scale inside agent platforms.
Whereas extremely technical groups can select to construct their very own infrastructure on high of the open-source code, NanoCo will promote managed, organization-wide deployments, taking up the burden of well being checks, integrations, and ongoing safety upkeep.
Widespread international adoption
The open-source adoption of NanoClaw has been staggering, crossing 250,000 downloads and nearing 29,000 GitHub stars since its debut. This ground-up momentum is solely chargeable for the surging enterprise demand.
“Numerous enterprise executives have informed us the identical factor,” Cohen said within the press launch. “They're operating NanoClaw personally, getting two and thrice extra performed, and asking easy methods to roll it out to their groups.”.
Maybe probably the most high-profile validation got here through the founders' latest journey to Singapore. The nation’s Overseas Minister, Dr. Vivian Balakrishnan, invited the NanoCo workforce to his workplace after publicly posting about his private use of NanoClaw. Balakrishnan described the agent as “getting smarter over time," referred to it as his "second mind," and said he wouldn't "dare swap it off".
Cohen put the platform's safety claims to the last word take a look at throughout a dwell convention demonstration in Singapore. He invited a crowd of 300 folks to talk concurrently together with his private agent, which was actively linked to his actual electronic mail and calendar.
Due to NanoClaw's zero-trust gateway structure, the agent safely rejected malicious makes an attempt to entry his inbox or delete current occasions, whereas efficiently permitting 12 attendees to guide legit espresso chats.
As AI shifts from a novelty device that solutions questions right into a digital workforce that autonomously executes duties, NanoCo AI is betting that verifiable safety would be the defining metric of success. By combining a clear open-source core with strict, infrastructure-level sandboxing, they aren’t simply promoting an assistant; they’re promoting the peace of thoughts required for enterprises to truly use one.