Kevin Jiang helped build Comet at Perplexity. Now he’s betting the next chapter of AI browsers lies beyond search and summaries. His new startup, Polar, is building a browser that can carry out hours of browser-based work with little human input, taking on the repetitive online tasks that consume much of the workday.
That vision has attracted investors. Polar announced Tuesday that it has raised $5.7 million in seed funding in a round led by Madrona, with participation from angel investors including former GitHub CEO Thomas Dohmke and Modal founder Erik Bernhardsson.
For years, web browsers have been the place where knowledge work happens. AI made them better at finding information, writing emails, and summarizing web pages. Polar is betting the next leap is a browser that does the work itself, clicking, typing, researching, and completing long, multi-step workflows across the web with minimal human input. The funding comes as startups and tech giants race to build AI agents that can move beyond chatbots and perform real work inside the software people use every day.
Polar’s approach is built around a simple idea. Most knowledge work already happens inside a browser. Employees spend hours switching between tabs, copying information, filling out forms, researching prospects, updating CRM systems, reviewing candidates, and completing repetitive workflows. Polar aims to take over those tasks by interacting with websites the same way a person would.
With $5.7 million in funding, Polar launches AI browser that completes hours-long tasks across the web
Instead of relying on custom integrations or APIs, the browser clicks buttons, types into forms, scrolls through pages, and logs into websites using the user’s own accounts. The goal is to let people hand off multi-step jobs that would normally demand constant attention.
That focus separates Polar from many earlier AI browsers, which centered on answering questions, summarizing web pages, or organizing tabs. Polar says its browser can execute entire workflows from start to finish, running for minutes or, in some cases, many hours without human intervention. According to the company, the system has completed tasks lasting more than fifteen hours.
The startup says customers have already run more than 4.5 million actions through the platform during its first seven months. Those actions span research, recruiting, sales, operations, marketing, and other knowledge-intensive work. Polar adds that its browser agent outperforms systems from OpenAI and Anthropic on browser-agent benchmarks, a sign of how competitive the race to build reliable AI agents has become.
“Almost all knowledge work runs through the browser, and until now nothing could do that work for you,” said Kevin Jiang, co-founder and CEO of Polar. “We built Polar to take the busywork off people’s plates, on the sites they already use, so they can spend their time on the parts of the job that need a person.”
The company was founded by Kevin Jiang, Vishaal Ram, and Howard Zhong, whose backgrounds include MIT, Perplexity, and quantitative investing firms. Their experience reflects a broader trend across AI startups, where teams with expertise in machine learning, infrastructure, and financial systems are applying those skills to workplace automation.
Investors see browsers as one of the largest untapped opportunities for AI agents. Software engineers already use coding assistants that write, test, and debug software with minimal supervision. Many office workers still spend a significant part of their day completing repetitive browser-based tasks that have changed little over the last decade.
“Engineers got a transformative moment from tools like Claude Code, and most knowledge workers are still waiting for theirs,” said Sabrina Albert of Madrona, who led the investment. “Their work lives in the browser, and Kevin, Vishaal, and Howard built the first agent that can operate there. That is why we led the round.”
The funding arrives during a surge of interest in AI agents. OpenAI, Anthropic, Google, Microsoft, and a growing list of startups are investing heavily in systems that can reason through tasks, interact with software, and complete work with limited supervision. Browser automation has become one of the industry’s most active battlegrounds since so much business activity still flows through web applications.
Polar is betting that users do not need another browser with AI features. They need one that behaves like a digital coworker. If that vision proves reliable at scale, the browser could shift from being a place where work happens to one where work gets done on its own.
