- Two brothers built a $12M AI startup, NanoCo, in just six weeks.
- The company’s AI agent, Claw, can navigate digital workflows like a human employee.
- Claw can parse tone, intent, and urgency and perform tasks in under 90 seconds.
- NanoCo has raised $12 million in seed funding from top AI-focused venture funds.
- The company had no formal product demo, sales team, or website at the time of the funding.
It was a humid night in Tel Aviv when Gavriel Cohen, sleeves rolled to the elbows, stared at a flickering terminal screen in his family’s converted garage. Lines of Python code scrolled past, intercut with error messages and the occasional burst of success—a server responding, an API authenticating, a task completed without human intervention. Around him, crumpled protein bar wrappers and empty soda cans formed a quiet testament to weeks of sleepless iteration. His younger brother Lazer, barefoot and pacing, muttered commands into a voice recorder, testing whether their creation—’Claw,’ an AI agent designed to navigate digital workflows like a human employee—could parse tone, intent, and urgency. When Claw autonomously filed an expense report, scheduled a follow-up meeting, and drafted a client email in less than 90 seconds, they knew something had changed. Not just for them—but for how work itself might be reimagined.
The $12 Million Launch
NanoCo, the company the brothers formally incorporated just six weeks after that breakthrough, has raised $12 million in seed funding from some of Silicon Valley’s most aggressive AI-focused venture funds, including Index Ventures and a consortium led by former OpenAI executives. What makes the raise extraordinary isn’t just the speed or the sum—it’s that the company had no formal product demo, no sales team, and no website at the time of the term sheet. Investors were given a 14-minute screen recording of Claw navigating a simulated corporate environment: extracting data from Slack, updating a Google Sheet, initiating a Zoom call, and even drafting a compliance memo for legal review. According to one partner at a participating fund, ‘We didn’t invest in a product. We invested in the inevitability of agents that work.’ Claw, unlike traditional automation tools, doesn’t rely on pre-built integrations or rigid workflows. It operates across apps using vision models and natural language reasoning, mimicking human behavior at the UI layer—effectively ‘clicking’ and ‘typing’ like a real employee.
From Chess Bots to Corporate Agents
The Cohen brothers’ path to AI automation began not in enterprise software, but in competitive chess. Gavriel, a former International Master, spent years building AI bots that could learn strategy from game logs, while Lazer, a self-taught programmer, reverse-engineered browser automation tools to scrape tournament data. Their first joint project, a predictive model for chess openings, caught the attention of a fintech startup in 2020, which hired them to adapt the system for automating customer support triage. That project revealed a broader truth: most corporate software is designed for humans, not machines, creating a bottleneck for automation. Existing tools like Zapier or UiPath require APIs and structured data, but in reality, employees toggle between apps, interpret ambiguous messages, and make judgment calls—tasks traditional bots can’t handle. The idea for Claw emerged from this gap: an AI agent that doesn’t need API access but instead learns by watching and doing, much like a new hire. By 2023, they had built a prototype that could complete multi-step tasks across Gmail, Notion, and Salesforce with 87% accuracy, according to internal benchmarks.
The Minds Behind the Machine
Gavriel, 31, is the strategist—the one who maps out workflows and defines the ‘theory of action’ behind each agent task. Lazer, 27, is the tinkerer, obsessed with reducing latency and improving the agent’s real-time decision-making. Together, they operate with an almost telepathic rhythm, shaped by years of coding side-by-side and debating AI ethics over late-night falafel. They reject the notion that Claw is simply a cost-cutting tool. ‘We’re not replacing people,’ Gavriel insists. ‘We’re replacing busywork.’ Their vision is more nuanced: a future where employees offload repetitive tasks—scheduling, data entry, status updates—to AI agents that work in the background, learning from their behavior. Investors have embraced this human-centric framing, seeing it as a counterpoint to dystopian narratives about AI-driven job loss. Still, the brothers remain wary of hype. They declined to name their AI ‘assistant’ or ‘copilot,’ opting instead for ‘Claw’—a deliberate choice, Lazer says, ‘to remind people this isn’t cute. It’s capable.’
Ripple Effects Across the Workforce
The implications of NanoCo’s approach extend far beyond startup lore. If AI agents can operate seamlessly across unstructured digital environments, entire categories of mid-level administrative work could be transformed. Early pilot programs with European logistics firms show Claw reducing invoice processing time by 72% and cutting internal coordination delays by over half. But the shift also raises urgent questions about job displacement, data privacy, and auditability. Because Claw interacts with apps via the user interface, its actions are not always logged in backend systems—creating potential compliance blind spots. The company is working with regulatory experts at Reuters to develop transparency layers that allow managers to trace every decision an agent makes. Meanwhile, enterprise clients are eager for solutions. At least three Fortune 500 companies are running private trials, particularly in procurement and HR, where repetitive workflows are common but integration across legacy systems remains a challenge.
The Bigger Picture
NanoCo’s rise reflects a broader pivot in the AI industry—from models that generate text to agents that take action. As Nature recently noted, the next frontier of AI isn’t just intelligence, but agency. The Cohens’ garage-built breakthrough suggests that this shift may come not from Big Tech labs, but from small, agile teams solving real-world friction points. Their success also underscores a growing investor belief: that the most valuable AI companies won’t just provide insight, but execution.
What comes next for NanoCo is expansion—both in capability and scrutiny. The brothers plan to open a U.S. office in Austin and are hiring safety researchers to audit Claw’s decision-making. They’re also exploring ‘agent teams’—swarms of AI workers that collaborate on complex projects. But they remain grounded in their original insight: that the future of work isn’t about humans versus machines, but about designing systems where each amplifies the other. In a world racing toward full automation, the Cohens are building something subtler: a bridge.
Source: Fortune