When a user asks a question that requires live web data, the agent calls Browse AI's Run Robot API with the appropriate robot ID and input parameters like the target URL. The robot executes the scrape and returns structured results. The agent then formats this data into a conversational response. You configure which robots are available to the agent in your Tars dashboard.

Yes. Browse AI robots must be created and trained in your Browse AI account first. You teach a robot what data to extract by pointing and clicking on a webpage. Once the robot is ready, Tars can trigger it via the API. Think of Tars as the conversational interface and Browse AI as the data extraction engine behind it.

Yes. The agent can initiate Browse AI's bulk task endpoint, which supports up to 50,000 URLs in a single run. It passes the list of URLs and parameters, then checks back for results. For very large jobs, the agent can poll task status and notify the user when the extraction is complete.

The agent checks the task status returned by Browse AI. If a task fails due to anti-bot protection, page changes, or timeout, the agent informs the user with the specific error reason and suggests alternatives like retrying with different parameters or checking if the target website has changed its structure.

No. Tars retrieves scraping results from Browse AI in real-time during conversations and uses them only to respond to the current query. The extracted data lives in your Browse AI account. Tars does not maintain a separate copy of your scraped datasets.

Yes. The agent can create Browse AI monitors with custom schedules using natural language. A user can say 'Monitor this page every hour on weekdays' and the agent translates that into Browse AI's scheduling format and creates the monitor automatically, complete with text and screenshot change notifications.

Browse AI's dashboard is designed for manual setup and review. Tars adds a conversational layer so anyone on your team can trigger robots, check results, and set up monitors by simply asking. Non-technical staff can access web scraping capabilities without learning the Browse AI interface, and results arrive in the same chat where questions are asked.

Tars requires your Browse AI API key, which you can generate from your Browse AI dashboard. The API key grants access to your robots, tasks, and monitors. You can restrict permissions by creating keys with limited scopes. No admin password is needed, and you can revoke access at any time from your Browse AI settings.