BrowserPilot

BrowserPilot is a Python browser automation agent for developers who want to describe web tasks in plain English. It controls Chrome through Selenium on your machine, or a remote browser through Selenium Grid. The software is open source under the MIT license, but its AI capabilities use OpenAI's cloud API and require an API key and internet access.

The agent can follow links, fill inputs, click buttons and scroll pages, including content inside iframes. It can also use AI to identify a page element from a description, extract information from web pages and summarize what it reads. Those capabilities suit tasks that combine browser interaction with interpreting page content, such as searching a site and collecting relevant text into a file.

Written instructions can include reusable functions, and BrowserPilot accepts saved tasks in JSON or YAML. It can save compiled instructions for reuse, reducing repeated calls to the model for the same task. Its default model is gpt-3.5-turbo.

Some programming knowledge helps: precise descriptions of page elements work better than casual requests, so familiarity with Selenium matters when deciding if it fits your workflow. BrowserPilot sends instructions and relevant page content to OpenAI for AI processing. It also executes model-generated Python code locally, a security tradeoff for anyone considering giving it access to a browser session.