Skip to content

AI that operates
your browser

60+ browser automation tools plus a built-in AI assistant — chat about any page, run quick actions from a command palette, let an agent automate multi-step tasks, and record flows for replay.

Architecture

How browser automation works

The JieGou browser extension connects to your console via the Model Context Protocol (MCP). When a recipe or workflow needs to interact with a browser-based tool, it sends a tool call through this connection — and the extension executes it in your browser.

1

Recipe or workflow calls a browser tool

Your automation includes a browser tool step — like "read the current Gmail thread" or "fill this Jira form."

2

Extension executes in your browser

The MCP client in the extension receives the tool call and executes it against the live web page, using platform-specific handlers for accurate data extraction.

3

Results flow back to the workflow

The tool result (page content, form state, screenshot) is returned to the recipe or workflow step, which continues with the data.

AI Assistant

How the AI assistant works

Open the side panel on any page. The extension extracts context, connects to the Console backend, and gives AI full awareness of what you're looking at.

1

Open the side panel and ask a question

The extension extracts page context (up to 8 KB) and detects the platform — Gmail, Slack, Salesforce, and 10+ more.

2

AI responds with full page awareness

Streaming responses from the Console backend with conversation history. Switch to Agent mode for multi-step tasks.

3

Record, replay, and share

Capture interactions as replayable flows. Add variables, adjust speed, and export as JSON for team collaboration.

Capabilities

What the extension can do

From clicking buttons to capturing network traffic, the extension gives AI a full toolkit for browser interaction.

Page Interaction

Click elements, fill forms, select dropdowns, check boxes, scroll, and navigate between pages.

Content Reading

Extract visible text, specific elements by CSS selector, page titles, URLs, and full DOM structure.

Screenshots

Capture the visible viewport or specific elements for visual verification or multimodal AI input.

Network Capture

Monitor and capture network requests and responses to understand data flowing between browser and backend.

Dedicated Automation Window

An isolated browser context for running automations in the background without interfering with your work.

In-Browser ML

On-device embeddings via WASM/ONNX runtime — process data locally without sending it to external servers.

AI Chat

Chat with AI about the page you're viewing. Page context is extracted automatically with platform-specific handlers for Gmail, Slack, Jira, and more.

Command Palette

Press Cmd+Shift+K for instant actions: screenshot, extract markdown, pull links and tables, summarize, draft replies, or run custom AI prompts.

Agentic Browsing

Switch to Agent mode and describe a task. The AI clicks, navigates, and fills forms with per-action approval — or bulk-approve for trusted sequences.

Flow Recording

Record browser interactions, save as replayable flows with variable substitution. Export as JSON, import from teammates, or upload to the Console.

Security

Your browser, your data

The extension requires explicit user authorization and only connects to JieGou's MCP server. All tool calls are authenticated and logged in the audit trail. The extension operates in your browser — data flows between your browser and the web applications you use. JieGou orchestrates the automation but doesn't see page content.

Explicit authorization required Authenticated tool calls Full audit logging BYOK data model

Install the browser extension

Available for Chrome and Firefox. Sign in with your JieGou account and browser tools appear in your recipes and workflows immediately.