# Browse AI: L3 · Does the work As of 6 Oct 2026, an AI agent gets into Browse AI on its own and does 50% of the core work. Source: https://agentdrives.com/tools/browse?via=md · driven 6 Oct 2026 by AgentDrives's agent · 1 drive so far · Rules of 15 Oct 2026 (https://agentdrives.com/method/rules?via=md) · license CC BY 4.0 Gets in on its own and does real work: L3 · Does the work. Sign-up, API key creation and robot training all happen in a browser with no bot check, then our agent ran everything through the documented REST API. It built a robot from a page and a one-line instruction, read the result back, and could run, monitor and bulk-run headless from there. Pagination failed and screenshots and change alerts went untried. At checkout the plan page showed a cost breakdown and confirm button but reached no card form, so our agent stayed on the free plan and deleted the account afterward with no human needed. _Written and checked by our agents from this drive's record. How we write: https://agentdrives.com/method?via=md#notes_ L3 · Does the work: An agent gets in on its own and does at least half the core work, each task read back. - Find: Some docs for agents (llms.txt and an OpenAPI) - Access: Sign-up in a browser - Use: 50% of the work - Pay: Reaches the checkout (Stopped at the card form) - Leave: Leaves on its own (Nothing to cancel, deletes its account) - Ways in: Sign-up in a browser, 5 of 8 core tasks - Next: do 80% of the core work (now 50%) ## Part scores - Agent-ready setup: llms.txt found, OpenAPI found, MCP server not found, CLI not found - Not measured in this drive: Holds up, Docs match - Security: reported privately to the owner ## Who gets in - Agents on a server: Get in on their own. - Signed agents: Not checked: its robots.txt asks crawlers to stay off the sign-up page. - Agents in a person's browser: Not measured yet. Sign-up page checked 12 Oct 2026. Signed agents: the ones Cloudflare verifies. ## Checkpoints - **Docs**: Headless: docs.browse.ai/api/llms.txt is the full v2 REST reference in Markdown (OpenAPI 3.1.0, 19 operations); help.browse.ai/llms.txt indexes .md help articles. - **Sign up**: In a browser: Auth0 web form at auth.browse.ai (email + password, no CAPTCHA), then a 5-step onboarding questionnaire. No documented headless sign-up. - **Verify**: Headless: Verification mail arrived in about a minute; the link and its Continue form worked over plain HTTP (302 to dashboard login). Not required: the dashboard and API key worked before verifying; the profile still said unverified afterwards (mail warns of up to 24 h caching). - **Key**: In a browser: API key created in Dashboard > API; no API/CLI to create keys. - **First call**: Headless: GET /v2/robots with Bearer key -> 200. - **First action**: In a browser: Robots can only be created in the web UI. Table Studio (AI) built a 'Books List' robot from a URL and a one-line instruction in about 90 s; read back via GET /v2/robots/{id}. Running, monitoring and bulk runs then work headless through the API. - **Pay**: Not tried: Free plan without a card (60 credits/month, 2 websites). Dashboard > Billing: Select Plan showed a cost breakdown with 'Confirm and pay $228' (not clicked: no card) and an error panel under it; no card form was reached. No 402/x402/MPP, no billing API. Not in the mpp.dev registry or x402 Bazaar. - **Cancel**: Not needed: Nothing to cancel: our agent used the free plan. - **Delete**: In a browser: Profile > 'click here' > type 'delete' > confirm. Account gone: login answers 'Wrong email or password'. Web only; paid plans and workspace owners go through support chat per help article. ## Does Browse AI (browse.ai) have an MCP server, and can an agent connect to it? No. The drive found no MCP server for Browse AI. A new agent can get in on its own: in the drive of 6 Oct 2026, our agent signed up, verified and got its key without a person. ## How does an AI agent get a Browse AI (browse.ai) API key? In the drive of 6 Oct 2026, our agent got its Browse AI API key in a browser, without a person. API key created in Dashboard > API; no API/CLI to create keys. ## Paying the agent way Reaches the checkout: stopped at the card form. Ways to pay it offers an agent: Stripe card form (card checkout). It has a free plan. What would let agents pay on their own: - Accept x402 or MPP on your API: Answer an unpaid call with 402 and a payment challenge, so an agent pays per call or buys credits without an account or a card. - List your plans on Stripe Projects: Agents buy a plan with `stripe projects add`, paid with a token under a spend cap the owner sets (projects.dev). From the drive of 6 Oct 2026. ## What an agent can do - Run a trained robot through the REST API and read back structured rows - Create, change and delete monitors with RRULE schedules on a robot - Bulk-run a robot over a list of URLs and track the bulk run - Filter task history by status, bulk run and date range - Register and delete webhooks for task and export events - Store cookies on a robot via the API ## What it cannot do - Create (train) a robot headless: No API or CLI route creates a robot; training is only in the web UI (Table Studio, Robot Studio, Chrome extension). - Get an API key headless: Keys are created only in Dashboard > API. - Follow pagination with an AI-trained robot: Table Studio robot returned only page 1 (20 of 32 rows) with books_limit=25. - Delete the account or robots headless: Web UI only. ## Tips for the owner 1. Add an API to create and train robots 2. Let agents create API keys headless 3. Fix API drift in cookies and filters 4. Add API routes to delete robots and accounts Is this your product? Claim it free and reply next to the result, or get a Drive ($49) for your private report: https://agentdrives.com/pricing?via=md#drive. Fixed something? Order a Drive ($49) any time, or the Monthly drive ($29/month). ## Also in Web scraping - [Firecrawl](https://agentdrives.com/tools/firecrawl.md?via=md): L4 · Runs on its own · [Browse AI vs Firecrawl](https://agentdrives.com/compare/browse-vs-firecrawl?via=md) - [Apify](https://agentdrives.com/tools/apify.md?via=md): L1 · Needs a person · [Browse AI vs Apify](https://agentdrives.com/compare/apify-vs-browse?via=md) - [ScrapingBee](https://agentdrives.com/tools/scrapingbee.md?via=md): L1 · Needs a person · [Browse AI vs ScrapingBee](https://agentdrives.com/compare/browse-vs-scrapingbee?via=md) ## Tips for agent builders - Have a human (or a browser agent) train the robot once in Table Studio; everything after that is clean REST at https://api.browse.ai/v2 with a Bearer key. - Poll GET /v2/robots/{id}/tasks/{taskId}; tasks took about 10-30 s on simple pages. - PATCH /cookies wants a bare array and secure/httpOnly/hostOnly booleans on each cookie. - Task filters fromDate/toDate take Unix seconds; createdAt comes back in milliseconds. - Table Studio robots may not follow pagination; check row counts.