How does Rover interact with a website's structure to perform actions?
Rover operates by reading the Document Object Model (DOM) of a website, which allows it to understand the page structure as rendered by the browser. This DOM-native approach ensures that Rover's understanding of the site is always current and accurate, enabling it to plan and execute actions precisely.
What is the primary differentiator between Rover and traditional RAG chatbots or vision agents?
Rover distinguishes itself by being DOM-native, meaning it directly interacts with the website's structure to perform sub-second actions with first-party embed precision. Unlike RAG chatbots that only answer questions or vision agents that rely on slow screenshot analysis and pixel guessing, Rover can actively click, input, validate, and navigate within the user interface.
Can Rover assist with user onboarding and guided tours within a product?
Yes, Rover is designed to guide users through product setup and feature discovery. It can click alongside users in real-time, leading to 60% faster activation and fewer drop-offs during the setup process, making it suitable for B2B onboarding, training, and workflows.
What is the typical latency for Rover to execute an action on a website?
Rover executes actions instantly with sub-second latency. It avoids the delays associated with screenshot loops or pixel-based analysis, ensuring native browser precision and a seamless user experience.
How is Rover integrated into a website, and what is required for its setup?
Rover is integrated by adding a single line of JavaScript code to a website. This one-line embed means there's no need for a knowledge base, embeddings pipeline, or extensive configuration; Rover reads the site live once the script is in place.