Repository navigation
ai browser vs ai browser agent
The two terms get used interchangeably and they name opposite things. An AI browser is a browser you use, with an AI built into the window: you do the browsing, it summarizes, answers, and increasingly offers to take the wheel for a task. An AI browser agent is software that does the browsing: you state a goal, it drives a browser - often one you never see - and returns the result. One is a better window; the other is a worker. Most disappointment in this category comes from buying one when the problem needed the other, so this page draws the line, lists what actually exists on each side as of September 2026, and gives you the test.
Disclosure, as on every comparison here: this wiki belongs to invisible_playwright_mcp, an open-source project on the agent side of the line. Product statuses below were verified against vendors' pages and coverage on 2026-09-03.
The category's living products, checked this session:
Comet (Perplexity). A Chromium-based browser with Perplexity's assistant built in, able to summarize, draft, and act on pages. It rolled out on Windows and macOS from July 2025, Android in November 2025, and iOS in March 2026, and has been free for everyone since late 2025. It is currently the most complete consumer AI browser: real platforms, no paywall, an assistant that can take actions in the page. Worth knowing when you weigh the category: researchers demonstrated a prompt injection exploit dubbed CometJacking, which Perplexity disputed and then patched - assistants that act on live pages inherit live-page risks, whichever vendor ships them.
Dia (The Browser Company, under Atlassian). Atlassian agreed to buy The Browser Company for $610 million in September 2025 with Dia as the focus. Dia today is in beta, macOS 14+ on Apple silicon only, and leans knowledge-work: briefs of your calendar and inbox, synthesis across tools, chat with your tabs. Design-forward and promising, but a Windows user cannot run it, and it is a workplace bet more than a general one.
Chrome with Gemini. The incumbent went AI browser in place: a Gemini side panel, and since 28 January 2026 the auto browse agent for AI Pro and Ultra subscribers in the US, which drives pages in a marked tab with an action log and confirmation gates. Chrome is the clearest evidence that the AI browser is becoming a feature of every browser rather than a product category.
ChatGPT's browsing surfaces. OpenAI tried a standalone AI browser - Atlas, launched October 2025 - and shut it down in August 2026, concluding, per coverage of the shutdown, that the browser is "a feature, not the destination". What remains is agentic browsing inside the ChatGPT desktop app and a Chrome extension. Claude in Chrome (generally available on paid Claude plans since 26 August 2026) takes the same extension-shaped route: your Chrome, with an AI that can act in it.
Read the pattern across those four: every AI browser is growing an agent mode, and one vendor already concluded the standalone AI browser was not worth shipping. The category is converging on "your browser, plus an assistant that sometimes acts".
On this side, the browser is a tool the software holds, not a window you look through. You interact with a prompt, a script, a schedule or an assistant; the agent interacts with the web.
The shapes it comes in:
- Hosted general agents - Manus, ChatGPT Work, Gemini Agent - take a goal and run it on vendor infrastructure; Manus alternatives maps that shelf and its ownership turbulence.
- Open-source agents you run - browser-use, Skyvern, Agent S, and our own invisible_playwright_mcp - live on your machine with your model key. The directory pages are Open-source AI browser agents and Open-source computer-use agents.
- API building blocks - the computer use tools in the Gemini and Claude APIs - for building your own; Gemini computer use vs Claude computer use referees those.
The agent side is what you want when the task is the point and watching it is not: extraction to a file, monitoring a page on a schedule, filling the same form forty times, research that spans thirty tabs you never want open. It is also the side that can run headless, on a server, from cron, or inside an MCP assistant - none of which a browser you browse with can do for you.
Auto browse in Chrome and Comet's assistant genuinely act on pages, so "browsers do not act" is already false. The distinction that still holds is structural:
- Whose session and identity does the work? An AI browser acts as you, in your logged-in profile, on your machine, while you watch. An agent runs its own browser - its own profile, possibly its own fingerprint and egress - and can do so without you present.
- What is the unit of use? A browser's unit is a browsing session you are having. An agent's unit is a task: it can be scripted, scheduled, parallelized and piped, which is why extraction, monitoring and research live on the agent side of this wiki.
- Who is accountable to the page? In an AI browser, sites see your normal browser with you behind it. An agent's browser answers a page's questions itself, which is where the engineering under agents diverges - what a site can tell, and what that costs, is its own section here.
- You browse all day and want it to hurt less - summaries, drafting, tab wrangling, an occasional "just do this bit for me": AI browser. Comet if you want it free and cross-platform today; Chrome if you want it inside what you already run; Dia if you are macOS and work-tool centric.
- You have tasks, and the browser is incidental to them: agent. If the tasks are occasional and you already pay OpenAI, Google or Anthropic, their built-in agents cover the easy cases. If the tasks recur, need scheduling, touch your own infrastructure, or you want the code, use the open-source shelf - start at Choosing an AI browser agent.
- You want both: they compose. A browser with an assistant for your hands-on hours, an agent for the work that should not need you. Nothing about the choice is exclusive except your budget.
The disclosure, once more, because this is the paragraph where it matters: we make an agent, so "you probably want an agent" is exactly what we would say. The test we actually stand behind cuts both ways - if you would watch it work, you wanted a browser; if watching it work would be a waste of your afternoon, you wanted an agent.
What is the difference between an AI browser and an AI browser agent? An AI browser is a browser you use with AI in it; an agent is software that operates a browser for you, often unattended.
Is Comet an AI browser or an agent? A browser - with an assistant that can act on the page you are on. It does not run unattended tasks on your behalf the way agent software does.
Is ChatGPT an AI browser? OpenAI's standalone browser, Atlas, shut down in August 2026; ChatGPT now offers agentic browsing inside its desktop app and a Chrome extension.
Are AI browsers safe? They inherit live-page risks: Comet's CometJacking prompt injection exploit was demonstrated and patched, and every vendor gates sensitive actions behind confirmations. The same category of risk applies to agents, where you choose the guardrails.
Do I need an AI browser to use an AI browser agent? No. Agents bring their own browser - invisible_playwright_mcp ships its own patched Firefox, others drive Chromium builds - and run alongside whatever you browse with.
Can one product do both? Chrome with auto browse is the closest today: a normal browser that can hand a tab to an agent. The unattended, scripted, scheduled cases remain agent territory.
See also: What is an AI web agent? for the agent category from first principles, Choosing an AI browser agent for picking one, and Manus alternatives for the hosted general agents that sit above both categories.
- Wikipedia: Comet (browser), fetched 2026-09-03: platform rollout dates, the free-for-everyone change, the assistant's capabilities, and the CometJacking episode.
- Dia's site, fetched 2026-09-03: beta status, macOS 14+ Apple silicon requirement, and feature descriptions. CNBC: Atlassian agrees to acquire The Browser Co. for $610 million, surfaced via search 2026-09-03.
- 9to5Google: Chrome rolling out Gemini 3-powered auto browse, fetched 2026-09-03: launch date, tiers, confirmation gates and the action log.
- TechCrunch: OpenAI is shutting down Atlas, fetched 2026-09-03: the shutdown, the "feature, not the destination" framing, and where the capabilities moved.
- Anthropic: Claude in Chrome is generally available, surfaced via search 2026-09-03, for the 26 August 2026 general availability on paid plans.
From the maintainers of invisible_playwright_mcp, an open-source AI browser agent - the side of this page's line we live on, which is why the browser side above is described entirely in its vendors' own verified terms.
- OpenAI Operator alternatives
- Open-source Operator-style agents
- Is OpenAI Operator still available?
- OpenAI Operator vs Claude computer use
- browser-use alternatives
- Choosing an AI browser agent
- Open-source AI browser agents
- Open-source computer-use agents
- What is an AI web agent?
- AI browser agents vs traditional scraping
- Cloud browser infrastructure for AI agents, explained
- Browserbase alternatives
- Firecrawl vs an AI browser agent
- Skyvern alternatives
- Stagehand vs browser-use
- Project Mariner is gone: what replaced it
- Manus alternatives
- Gemini computer use vs Claude computer use
- A stealth browser MCP, reviewed honestly by its own wiki
- AI browser vs AI browser agent: which one do you want?
- AI browser agent vs RPA: which one fits the job
- AI browser agent vs n8n, Zapier and Make
- Vercel agent-browser alternatives, compared honestly
- What is an agentic browser? Definition and the two kinds
- Open-source agentic browsers: the three layers, compared
- Choosing an MCP server for browser automation: four axes
- Stealth MCP servers compared: Camoufox, nodriver, Patchright
- Playwright MCP alternatives, and the three you don't need
- Autonomous browser agents: the four rungs of autonomy
- What is actually free in the AI browser agent stack
- browser-use on GitHub: what the repo actually gives you
- Playwright MCP vs Chrome DevTools MCP: different jobs
- How to choose among MCP servers: a map by category
- Which MCP servers are worth adding to Claude Code
- MCP on GitHub: finding servers and judging them fast
- MCP vs an API: the decision, and what the wrapper costs
- MCP alternatives: when the protocol is the wrong shape
- Why does my AI agent get blocked?
- The timing signal AI agents give off
- Agent retry loops trip rate limits, not fingerprints
- Claude computer use detected as a bot
- browser-use getting blocked: what you can and cannot change
- Playwright MCP session blocked: four causes, four fixes
- Playwright MCP and captchas: what actually gets you past
- Cloudflare and a browser MCP server: what is being read
- Can an AI agent solve a captcha? The honest answer
- Getting an AI agent to fill out forms
- The best model for an MCP browser agent, and what it really costs
- Browser problem or model problem?
- How to use a Playwright MCP server with Claude Code
- Extracting data to a CSV with an AI agent
- Monitoring a page for changes with an AI agent
- How to let Claude Desktop control a browser
- How to add a browser to Cursor as an MCP server
- Using an AI agent to hunt for apartments
- Getting website data into Google Sheets with an AI agent
- Using an AI agent to download invoices from portals
- AI agents for web research
- Using an AI agent to test your own website
- How to add a browser to Cline as an MCP server
- Posting to social media with an AI agent
- Posting to Facebook with an AI agent
- Posting to Instagram with an AI agent
- Posting to X with an AI agent
- Automating LinkedIn posts: read this first
- Appointment bots: what they are and what an agent can legitimately do
- Track prices across sites with an AI agent
- Build a lead list with an AI browser agent
- Run an AI browser agent on a schedule
- AI browser agent with a local LLM: what changes
- Should you log your AI agent into your accounts?
- How to write a task an AI browser agent can follow
- Move data between two web apps with an AI agent
- The MCP server
- How the tools are shaped, and why
- Playwright MCP vs the Playwright CLI: which fits when
- Playwright MCP: browser is already in use, and the fix
- Playwright MCP best practices: four decisions that matter
- Playwright MCP with a proxy, and the three leaks it leaves
- A browser MCP server in GitHub Copilot: setup and limits
- Using a browser MCP server for web scraping: the pattern
- Which LLM for browser automation: the four properties
- How to build a browser agent, and what to take instead
- Getting an AI agent to log into a website: three routes
- MCP tools, resources and prompts: who controls each
- How many MCP tools is too many? The context arithmetic
- How to build an MCP server: the decisions, not the scaffold
- Local or remote MCP server: what changes, and what does not
- Writing an MCP client in Python: the thirty-line version
- Self-hosted AI agent: what one actually costs to run
- How long an AI browser agent takes per step, measured
- Text, HTML, snapshot or screenshot: what the agent should read
- Giving an AI browser agent a stopping condition
- Keeping an AI browser agent out of destructive actions
- Why did the AI agent click the wrong thing
- When the page changes under the AI agent
- Running one AI agent task across a list of sites
- Seeing a page as it appears in another country
- Getting data out of a dashboard with no export button
- Two browsers in one session: main and support
- Finding the dead links on a site with an AI agent
- Filling a CRM record from a company's website
- One form submission per spreadsheet row, with an AI agent
- Dated screenshots of a page as evidence
- Checking order and delivery status with an AI agent
- Reading a PDF that opens inside the browser
- Summarising a long page or thread with an AI agent
- Collecting every image on a page with its caption
- Collecting event and course listings with an AI agent
- Cancelling a subscription with an AI agent
- What an AI agent can and cannot do inside an iframe
- Shadow DOM and an AI agent: you can click it, you cannot read it
- How to add a browser to Codex as an MCP server
- What a page snapshot costs, per control
- How to let Gemini CLI use a browser
- Native selects and the ones that only look like selects
- Clicking by selector or by coordinates
- How long the agent waits before it gives up
- What a second browser costs
- Uploading a file with an AI agent, and why this one cannot
- Watching the agent work, and when it is worth it
- When not to use an AI browser agent
- Agent or script: deciding once instead of every time
- Using the keyboard instead of the mouse
- Secrets in an agent task: where they end up
- What an agent run should log
- Deduplicating what an AI agent collects
- Normalising values across sites
- Validating an AI agent's output
- Reading a table with an AI agent
- Driving a site's own search and filters
- The task works headed and fails headless
- How to let the AdaL CLI use a browser