Dia Browser Hands-On Review — What the Arc Team's AI Browser Actually Changed
The biggest change in Dia browser isn't a "built-in AI chat window" — it's turning tabs themselves into the AI's context source. The 12 tabs you have open are p
The biggest change in Dia browser isn't a "built-in AI chat window" — it's turning tabs themselves into the AI's context source. The 12 tabs you have open are part of the prompt, with no copy-pasting required. This design compresses "asking questions across tabs" from 3-5 manual steps down to 1, and it's the core reason the Arc team halted major Arc development in 2025 and pivoted entirely to Dia. What Dia Is: The Arc Team's Second Bet Dia is developed by The Browser Company of New York, first released in June 2025 as an invite-only beta for macOS, then opened to public download in September 2025. The company's previous product was the Arc browser, which built significant buzz in the Mac user community in 2024, but "Arc's feature adoption stalled, with most users only using two features: the sidebar and Spaces" (source: The Browser Company official open letter, May 2025) . CEO Josh Miller accordingly announced that Arc would move to a security-updates-only maintenance state. In September 2025, Atlassian announced its acquisition of The Browser Company for $610 million, making Dia part of Atlassian's product lineup. "Atlassian acquires The Browser Company for $610 million in all cash" (source: Atlassian official announcement, September 2025) — the significance of this deal is that Dia went from an independent team's experiment to a product backed by enterprise software distribution channels. Under the hood, Dia is built on Chromium just like Arc, so Chrome extensions, password managers, and developer tools are all compatible. This is no small thing: the Chromium foundation lets Dia sidestep the "new browser with no ecosystem" death trap. The Core Mechanism: Tabs as Context Dia's differentiating feature centers on one design decision: by default, the AI can read the contents of every currently open tab. The traditional approach is for the user to copy the page text, paste it into ChatGPT, then type the question. With Dia, you press Cmd + E to bring up the chat box and
FAQ
What Dia Is: The Arc Team's Second Bet
Dia is developed by The Browser Company of New York, first released in June 2025 as an invite-only beta for macOS, then opened to public download in September 2025. The company's previous product was the Arc browser, which built significant buzz in the Mac user community in 2024, but "Arc's feature adoption stalled, with most users only using two features: the sidebar and Spaces" (source: The Browser Company official open letter, May 2025) . CEO Josh Miller accordingly announced that Arc would m
Hands-On Numbers: Which Scenarios Actually Save Time
The following comes from two consecutive weeks of daily use, tested on an M2 MacBook Air with an average of 15 to 25 tabs open at once. Cross-tab comparison : Comparing plan differences across 4 SaaS pricing pages took about 8 minutes with the traditional workflow (open tabs, read each page, take notes manually, organize into a table); with Dia, a single question produced a comparison in about 40 seconds, though 1 to 2 minutes of manual number-checking was still needed. Long-form summarization :
How It Actually Differs from Competitors
The AI browser race got crowded fast in the second half of 2025, with the main rivals being Perplexity's Comet and OpenAI's Atlas. The three have distinctly different design philosophies. Comet puts its emphasis on "agentic execution" — it actually operates web pages, fills out forms, and places orders. Perplexity's official positioning statement for Comet emphasizes agent capability, which suits process automation but has consequently drawn more security scrutiny. Atlas was launched by OpenAI i
Who Should Use It, Who Shouldn't
Dia's benefits are most obvious for work styles that require "synthesizing information across multiple sources every day": market research, competitive analysis, technology selection evaluation, initial academic literature screening. The bottleneck in this kind of work is reading and integration, not operation. The unsuitable cases are equally clear: work centered on a single application (design, video editing, software development), needing Windows or Linux support, belonging to an organization
Related Guidebooks
Reviewed and verified by FeiYueh · Last verified 2026-09-07. Independently maintained — not AI-generated boilerplate.
← Back to Blog