Polar is an AI browser aimed at knowledge workers, built to take repetitive web tasks off people's plates rather than reinvent consumer search. The platform clicks, types, navigates pages, and completes tasks while operating within websites where users are already authenticated , which lets it handle work across the logged-in tools a team already runs on. The launch matters now because the first wave of AI browsers leaned on chat-with-your-tab flows for consumers, and Polar is positioning around the harder problem of reliable agentic work inside real business systems.
The company was founded by Kevin Jiang, Vishaal Ram, and Howard Zhong , with CEO Kevin Jiang having previously worked at Perplexity on Comet . The launch coincides with a $5.7 million seed round led by Madrona , alongside participation from ex-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 .
Inside the browser, users can schedule workflows, save prompts and assign tasks to agents based on their open tabs , targeting use cases across sales, recruiting, marketing, research, and operations. Pricing is straightforward, with the browser itself free to use with a few daily credits for AI tasks, and subscription plans starting at $20 per month for higher limits. Polar reports that most current users run its browser alongside another daily driver purely to automate tasks , a useful signal for operators evaluating where agentic tooling actually sticks in day-to-day work.
Comments (14)
Sign in to join the discussion.
R
Renata VlkovaJul 29, 2026
Beating Anthropic and OpenAI on every major benchmark is a spicy first line for a launch tweet. Bold move to put the target on your own back before breakfast.
K
Kwame O.Jul 29, 2026
also the MIT flex in tweet 3 feels obligatory at this point, every AI browser founder is legally required to mention it.
K
Kwame O.Jul 29, 2026
The tweet copy is doing a lot of work but I want to see the actual demo before I believe the Clay cancellation story. Show me the workflow, not the receipt.
J
June ParkJul 29, 2026
naive q but what happens when the agent hits a captcha in the middle of automating recruiting outreach? does it just sit there forever?
D
Diego M.Jul 29, 2026
Whoever cut the launch video understood assignment on pacing but the hook is buried. First 2 seconds should be the agent doing a task, not a logo swirl.
S
Sanaa IqbalJul 29, 2026
hot take: 4.5M actions is a vanity metric unless we know how many were retries. I want a success rate, not a step counter.
O
Ola M.Jul 29, 2026
The browser wars are back and this time nobody wants to actually browse. Peak 2025 energy.
T
Tobias L.Jul 29, 2026
Fifth AI browser launch I've clocked this quarter and the pitch is converging fast. Whoever nails the enterprise trust layer wins, not whoever tops WebArena.
Y
Yuki TanakaJul 29, 2026
Is there an API or is this strictly a desktop client thing? Asking because rate limits on agent actions are where every one of these products quietly falls over.
P
Priyanka R.Jul 29, 2026
Curious which benchmarks specifically. WebArena, VisualWebArena, Mind2Web all measure very different things and 'every major' is doing a lot of lifting.
Y
Yuki TanakaJul 29, 2026
also webhooks when the agent finishes a task would be table stakes for the ops use case they keep advertising.
B
Bela HorvathJul 29, 2026
Every agent action costs inference. Curious what the pricing model looks like when a customer's SDR runs 500 workflows a day and the margin quietly evaporates.
C
chainmaxiJul 29, 2026
an agent that acts on your behalf is basically a wallet with hands. surprised there's no onchain identity angle here yet.
A
Amara NwosuJul 29, 2026
ok wait, if one company really canned Clay and saved 25 hrs/wk per person, that's the entire case study. Why is it buried in bullet three?
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Polar Browser
The AI browser that actually does work for you
The AI browser that actually does work for you
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Polar is an AI browser aimed at knowledge workers, built to take repetitive web tasks off people's plates rather than reinvent consumer search. The platform clicks, types, navigates pages, and completes tasks while operating within websites where users are already authenticated , which lets it handle work across the logged-in tools a team already runs on. The launch matters now because the first wave of AI browsers leaned on chat-with-your-tab flows for consumers, and Polar is positioning around the harder problem of reliable agentic work inside real business systems. The company was founded by Kevin Jiang, Vishaal Ram, and Howard Zhong , with CEO Kevin Jiang having previously worked at Perplexity on Comet . The launch coincides with a $5.7 million seed round led by Madrona , alongside participation from ex-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 . Inside the browser, users can schedule workflows, save prompts and assign tasks to agents based on their open tabs , targeting use cases across sales, recruiting, marketing, research, and operations. Pricing is straightforward, with the browser itself free to use with a few daily credits for AI tasks, and subscription plans starting at $20 per month for higher limits. Polar reports that most current users run its browser alongside another daily driver purely to automate tasks , a useful signal for operators evaluating where agentic tooling actually sticks in day-to-day work.
Comments (14)
Beating Anthropic and OpenAI on every major benchmark is a spicy first line for a launch tweet. Bold move to put the target on your own back before breakfast.
also the MIT flex in tweet 3 feels obligatory at this point, every AI browser founder is legally required to mention it.
The tweet copy is doing a lot of work but I want to see the actual demo before I believe the Clay cancellation story. Show me the workflow, not the receipt.
naive q but what happens when the agent hits a captcha in the middle of automating recruiting outreach? does it just sit there forever?
Whoever cut the launch video understood assignment on pacing but the hook is buried. First 2 seconds should be the agent doing a task, not a logo swirl.
hot take: 4.5M actions is a vanity metric unless we know how many were retries. I want a success rate, not a step counter.
The browser wars are back and this time nobody wants to actually browse. Peak 2025 energy.
Fifth AI browser launch I've clocked this quarter and the pitch is converging fast. Whoever nails the enterprise trust layer wins, not whoever tops WebArena.
Is there an API or is this strictly a desktop client thing? Asking because rate limits on agent actions are where every one of these products quietly falls over.
Curious which benchmarks specifically. WebArena, VisualWebArena, Mind2Web all measure very different things and 'every major' is doing a lot of lifting.
also webhooks when the agent finishes a task would be table stakes for the ops use case they keep advertising.
Every agent action costs inference. Curious what the pricing model looks like when a customer's SDR runs 500 workflows a day and the margin quietly evaporates.
an agent that acts on your behalf is basically a wallet with hands. surprised there's no onchain identity angle here yet.
ok wait, if one company really canned Clay and saved 25 hrs/wk per person, that's the entire case study. Why is it buried in bullet three?