Donut Labs Raises $22M for AI Crypto Trading Browser

The $22 million funding round, spanning pre-seed and seed stages, was led by an impressive consortium of venture capital firms including Bitkraft Ventures, Sequoia China, Sky9 Ventures, Matrix Partners, Makers Fund, and Altos Ventures. This substantial backing positions Donut Labs among the early pioneers racing to redefine how artificial intelligence and cryptocurrency markets converge. The company’s vision emerged from founder Chris Zhu’s frustration with existing platforms after his team’s TikTok-based app was abruptly shut down, leading to a fundamental realization about the future of online interfaces.

Zhu explained the core insight driving Donut’s development: ‘Browsers haven’t really changed in 30 years. We wanted to build the largest front end on the internet—but do that for crypto.’ He observed the emerging trend of agentic browsers, citing OpenAI’s Atlas and Perplexity’s Comet as early indicators of the shift. ‘We saw signs that this agentic browser wave was starting, and we wanted to create one that could actually handle trading,’ Zhu told Decrypt during the announcement.

Donut’s AI browser is engineered to understand a user’s complete trading context, synthesize data from multiple sources, and perform actions across exchanges—all within the browser environment. The system employs a sophisticated layered security architecture modeled after the ‘levels’ used in Tesla’s self-driving software. This approach creates a graduated security framework where tasks range from pure research to fully authorized transactions requiring multiple confirmations, ensuring appropriate safeguards for different risk levels.

Security vulnerabilities common in other AI systems were a primary design consideration. ‘The reason prompt injection happens in other browsers is because they handle HTML links with not enough security,’ Zhu explained. To address this critical concern, Donut processes authorizations on the backend rather than simulating user clicks, fundamentally changing how security is implemented. The company has partnered with key management firm Turnkey to safeguard private keys, with Zhu emphasizing that ‘we actually don’t give the agent the ability to click into your other wallets. Our wallet backend operates separately so the agent never sees your balances.’

Behind the scenes, Donut employs an ‘open router’ system that dynamically selects the most appropriate AI model for each specific task. This intelligent routing ranges from lightweight models for straightforward token swaps to advanced reasoning systems for complex market analysis. The browser’s agents leverage liquidity and pricing data from established sources including CoinGecko and Pyth, while custom-built tools enable automated execution and comprehensive research capabilities.

The company plans to deploy a complete product ecosystem covering what Zhu describes as ‘the full life cycle of your trading career.’ This comprehensive approach includes a standalone browser, browser extension, web application, and mobile application, ensuring traders can maintain their AI-powered capabilities across all devices and contexts. The multi-platform strategy reflects Donut’s ambition to become the primary interface for crypto trading activity.

Contrary to concerns about AI rendering human traders obsolete, Zhu presents a compelling vision of AI as a trading equalizer. ‘Right now, quants are great at high-frequency trades,’ he noted, ‘but AI browsers know your whole context—they make it easier for you to do what you already do, across the internet.’ This perspective suggests that rather than replacing human decision-making, AI browsers will amplify individual trading strategies and capabilities.

Zhu envisions a future where personalized AI assistance becomes the norm: ‘Over a long enough time horizon, we do want to get to the point where it’s just me, myself, and my hundred quants. But the strategies will be unique to each person.’ As AI systems evolve from conversational chatbots to fully autonomous tools, browsers may become the primary interface for both human and machine activity online. Zhu believes crypto traders, already accustomed to automation and market volatility, will be among the first to embrace this technological shift, noting that ‘our species is looking for rapid growth. Choosing AI to help us get there is one of the easiest ways to evolve.’