World's first agentic workflow exchange
rtrvr.ai empowers builders to automate their most complex web workflows and scraping using dead simple natural language and the SOTA AI Web Agent, via Web, API, Browser Extension, or Whatsapp
Retriever's Knowledge Base (RAG) feature indexes open tabs or scraped URLs using Gemini File Search. The agent can query the KB mid-workflow for smarter decisions, and users can chat with indexed documents for cited answers.
Retriever supports WhatsApp commands to trigger workflows remotely, MCP endpoints to give ChatGPT/Claude/Cursor browser control, and CLI/SDK integration for developers to route tasks between cloud and local extension.
Retriever offers scheduled automation that runs workflows hourly, daily, or on custom cron schedules. Results auto-sync to Google Sheets with timestamps, and the agent can access protected data behind logins using the user's active browser session.
Retriever turns the browser into a self-driving AI agent that accepts natural language commands to automate clicking, typing, navigation, and data extraction. Users can teach it tasks once by performing them manually, and it records actions and API calls to create reusable workflows.
Retriever uses DOM-based scraping for precision data extraction from thousands of rows. Results export to Google Sheets or structured JSON instantly, with higher accuracy than vision-only agents.