Sentience SDK is a browser automation framework designed for AI agent developers. Instead of using brittle CSS selectors, you find elements by their semantic meaning (role, text, importance).
What is the Sentience SDK?
Traditional browser automation relies on CSS selectors that break when websites redesign. Vision models are expensive and non-deterministic, requiring multiple retries. Sentience SDK provides a third way: semantic element discovery with deterministic execution.
Key Benefits
Semantic Queries: Find buttons by text, not CSS classes
Smart Ranking: Elements sorted by importance (what users actually interact with)
Realistic Clicks: Uses real mouse simulation, not just JavaScript events
Smart Waiting: Polls for elements using semantic conditions
Developer Tools: Inspector, Recorder, and Script Generator
How It Works
Extension: A Chrome extension collects page data (elements, positions, roles)
SDK: Your code calls SDK functions to interact with the page
Server API (Pro/Builder/Teams/Enterprise): Advanced ranking and filtering happens on our servers
Local Processing (Free): Basic data collection happens in your browser
Architecture
Key Features
Deterministic by Design
Same input yields same output, every time. No vision model guessing or random failures.
No Retry Tax
Vision-based agents often incur costs on every retry. Semantic geometry reduces retries by providing stable, deterministic targets.
Semantic Element Discovery
Find elements by role and meaning, not brittle CSS selectors that break on redesigns.
Production Ready
Built-in waiting, visibility checks, and error handling for reliable automation.
Developer Tools
Inspector, Recorder, and Script Generator to speed up development.