{"slug": "kevin-jiang-s-polar-raises-5-7-million-to-automate-knowledge-work-in-browsers", "title": "Kevin Jiang's Polar raises $5.7 million to automate knowledge work in browsers", "summary": "Kevin Jiang, a former Perplexity employee who worked on the Comet browser, launched Polar, an AI browser for knowledge work, and raised a $5.7 million seed round led by Madrona. Polar operates inside authenticated browser sessions, saves reusable workflows, and targets research, recruiting, sales, marketing, and operations. The browser runs on macOS and Windows 10 or later, with pricing from free to $100 per month.", "body_md": "On July 29, [Kevin Jiang](https://www.linkedin.com/in/kevinjiang7?ref=runtimewire), who previously worked on Perplexity's Comet browser, launched [Polar](https://polarbrowser.com/?ref=runtimewire) as an AI browser for knowledge work and raised a $5.7 million seed round led by [Madrona](https://www.madrona.com/?ref=runtimewire), [TechCrunch reported](https://techcrunch.com/2026/07/29/perplexity-employee-who-worked-on-comet-launches-an-ai-browser-aimed-at-knowledge-work/?ref=runtimewire).\n\nJiang is carrying a specific lesson out of Perplexity: an AI browser needs frequent, consequential work rather than occasional consumer errands. He attended MIT from 2018 to 2023, where his public profile lists four appearances on the USA Taekwondo National Team and a stint as co-president of the MIT Poker Club.\n\nHis diagnosis is that early AI browser demos centered on tasks people rarely repeat. Booking a flight or making a restaurant reservation can demonstrate an agent's ability to navigate the web, but those errands do not create a daily habit. \"We're focused on where we think browser agents are actually valuable, which is knowledge work,\" Jiang [told TechCrunch](https://techcrunch.com/2026/07/29/perplexity-employee-who-worked-on-comet-launches-an-ai-browser-aimed-at-knowledge-work/?ref=runtimewire).\n\nThat distinction shapes Polar's product. Polar can operate inside a user's authenticated browser sessions, accept tasks from an open tab, save prompts as reusable workflows and run them on hourly, daily, weekly or custom schedules. Polar is targeting research, recruiting, sales, marketing and operations, where workers repeatedly gather information and move it among web applications.\n\n### A browser built around delegated work\n\nPolar began as Composer, which [entered alpha on January 26](https://polarbrowser.com/blog/introducing-composer?ref=runtimewire). Composer used an Electron application and a cloud-based virtual file system to coordinate agents working across longer tasks. Polar later acknowledged that the architecture made it difficult to maintain authenticated sessions, interact with complex interfaces, support extensions and build persistent context around a user's work.\n\nPolar [announced the transition from Composer to Polar](https://polarbrowser.com/blog/new-interface?ref=runtimewire) on May 7, with the new product scheduled to arrive on May 18. Polar replaced the Electron approach with a Chromium fork, giving Polar deeper control over how the agent observes pages, handles sessions and interacts with web applications.\n\nThat technical shift supports Jiang's broader bet. A standalone chatbot can suggest what a user should do next. A browser already contains the user's logged-in accounts, open tabs and working context, allowing an agent to carry out the task directly. Polar says users can watch the agent, interrupt it and take over when judgment or sensitive action is required.\n\nPolar's current website says the browser runs on macOS and Windows 10 or later, with the Windows version in beta. That is an expansion from Polar's May announcement, which described the new browser as macOS-only. The [pricing structure](https://polarbrowser.com/pricing?ref=runtimewire) includes a free tier with daily credits, a $20-per-month Pro plan and a $100-per-month Max plan. Longer agent tasks consume more credits.\n\nPolar does not require users to abandon their existing browser, according to [Polar's website](https://polarbrowser.com/?ref=runtimewire). That gives Polar a narrower entry point: it can prove useful as a work agent before asking users to migrate the rest of their browsing activity.\n\n### The browser is the permission layer\n\nThe same access that makes browser agents useful creates Polar's hardest product problem. An agent operating inside authenticated sessions may encounter private email, internal systems, customer records and financial information. It must also interpret untrusted material from websites without treating malicious text as a user command.\n\n[Brave demonstrated the risk](https://brave.com/blog/comet-prompt-injection/?ref=runtimewire) when its researchers described an indirect prompt-injection attack against Perplexity Comet. The report underscored a security challenge for the entire agentic browser category: agents need to distinguish a user's instructions from untrusted text on the web.\n\nPolar says it has guardrails intended to stop the agent from independently taking high-risk actions and gives users the ability to observe or interrupt its work. Polar's website lists SOC 2 compliance as in progress. Turning those controls into an enterprise-ready security model will matter as Polar moves from research and prospecting tasks toward workflows that can change records, contact customers or send messages.\n\nAtlassian has already placed a larger bet on the same market. It [announced a definitive agreement to acquire The Browser Company](https://www.atlassian.com/blog/announcements/atlassian-acquires-the-browser-company?ref=runtimewire) on September 4, 2025, then [completed the acquisition on October 20, 2025](https://www.sec.gov/Archives/edgar/data/1650372/000165037226000011/R15.htm?ref=runtimewire) for a total purchase price of $488.3 million. The deal gave Atlassian Dia, a browser positioned around knowledge workers using SaaS applications. Polar enters with a narrower automation pitch.\n\n### What the seed round has to prove\n\nMadrona led Polar's $5.7 million seed round, and [Madrona partner Sabrina Albert](https://www.madrona.com/team-profiles/sabrina-albert/?ref=runtimewire) is the named partner associated with the financing. The investor group also includes former GitHub CEO Thomas Dohmke, Phia founder Phoebe Gates, Modal CEO Erik Bernhardsson, Flapping Airplanes co-founder Benjamin Spector, Etched co-founder Rob Wachen and Fundamental Research Labs co-founder and CEO Robert Yang.\n\nThe lineup brings operators from developer infrastructure, AI hardware and consumer software into Jiang's effort. It also reflects what Polar must build next: reliable long-running agents, a product non-technical employees can operate and security controls strong enough for work accounts.\n\nPolar has not published revenue, retention or user figures alongside the round. Its public case rests on the product's ability to complete recurring browser work and on [testimonials published on Polar's own website](https://polarbrowser.com/?ref=runtimewire). The seed financing gives Jiang room to turn those early examples into measurable usage across teams.\n\nJiang's timing rests on model capability as much as browser design. Long-running web agents depend on models that can retain a plan, recover from interface changes and recognize when a task requires human judgment. Better models make ambitious browser automation possible, while each additional action also creates another point where the agent can fail.\n\nPolar is therefore selling reliability rather than novelty. Polar can work without persuading every consumer to replace Chrome. The near-term test is whether recruiters, salespeople, researchers and operators trust Polar with work they perform every week, then every day. If that habit forms, the browser becomes less of a destination and more of an execution layer for the tools workers already use.", "url": "https://wpnews.pro/news/kevin-jiang-s-polar-raises-5-7-million-to-automate-knowledge-work-in-browsers", "canonical_source": "https://runtimewire.com/article/kevin-jiang-polar-raises-5-7-million-ai-browser-knowledge-work", "published_at": "2026-07-29 15:53:36+00:00", "updated_at": "2026-07-29 15:59:27.478794+00:00", "lang": "en", "topics": ["ai-products", "ai-agents", "ai-startups"], "entities": ["Kevin Jiang", "Polar", "Madrona", "Perplexity", "Comet", "TechCrunch", "MIT"], "alternates": {"html": "https://wpnews.pro/news/kevin-jiang-s-polar-raises-5-7-million-to-automate-knowledge-work-in-browsers", "markdown": "https://wpnews.pro/news/kevin-jiang-s-polar-raises-5-7-million-to-automate-knowledge-work-in-browsers.md", "text": "https://wpnews.pro/news/kevin-jiang-s-polar-raises-5-7-million-to-automate-knowledge-work-in-browsers.txt", "jsonld": "https://wpnews.pro/news/kevin-jiang-s-polar-raises-5-7-million-to-automate-knowledge-work-in-browsers.jsonld"}}