Browser Extension Configuration

The Browser Extension interface allows administrators to configure AI agent capabilities that users can access directly from their web browsers. Set up standard actions and create custom workflows that integrate seamlessly into your users’ browsing experience.

​

Browser Extension Download

​

Overview

Browser extension capabilities provide context-aware AI assistance that users can invoke while browsing. When configured, these capabilities appear as Quick Actions in the browser extension’s chat sidebar, allowing users to interact with your agents without leaving their current webpage.

​

Accessing Browser Extension Settings

1. Navigate to Settings in the Airia platform
2. Select Extensions from the sidebar
3. Click on Chrome Extension to view and manage capabilities

​

Standard Capabilities

The browser extension includes three built-in capabilities that you can enable and configure:

​

Summarize

Generates concise summaries of webpage content, helping users quickly understand long articles, documentation, or reports. Configuration:

• Agent Assignment: Select which agent handles summarization requests
• Context: Automatically includes the current page content

​

Key Points

Extracts main points from webpages in bullet format, ideal for research and information gathering. Configuration:

• Agent Assignment: Select which agent extracts key points
• Context: Automatically includes the current page content

​

Translate

Translates webpage content into different languages based on user requests. Configuration:

• Agent Assignment: Select which agent handles translation requests
• Context: Automatically includes the current page content

​

Configuring Standard Capabilities

For each standard capability:

1. Click Assign Agent or Change Agent to select an agent
2. Choose from your available agents in the dropdown
3. Click View Agent to review the agent’s configuration in a new tab
4. All changes are saved together when you click Save Changes

​

Custom Capabilities

Create custom capabilities tailored to your organization’s specific workflows and use cases. Custom capabilities appear alongside standard capabilities in the Quick Actions menu.

​

Creating a Custom Capability

1. Click Add Custom Capability in the Custom Capabilities section
2. Configure the following settings:

​

Display Name

The name users see in the Quick Actions menu. Choose a clear, action-oriented name (e.g., “Analyze Competitors”, “Extract Contact Info”).

​

Agent Selection

Select which agent powers this capability. The agent should be configured to handle the type of task this capability performs.

​

Mode

Choose how the capability accesses webpage content:

• Page: Includes the entire page content as context
• Selection: Only includes text the user has selected on the page
• Page Source: Provides the raw HTML source of the page

​

Initial Message

The prompt sent to the agent when users invoke this capability. This message:

• Can include instructions for how to process the content
• Should be clear about what the agent should do
• Can reference the page context that will be automatically included

Example Initial Messages:

• “Analyze this webpage and identify key competitors mentioned”
• “Extract all contact information from this page”
• “Summarize the technical specifications in a comparison table”

3. Click Add to create the capability

​

Editing Custom Capabilities

1. Click the Edit icon next to any custom capability
2. Modify the settings as needed
3. Click Save to apply changes

​

Deleting Custom Capabilities

1. Click the Delete icon next to the capability you want to remove
2. Confirm deletion in the dialog
3. The capability will be removed after clicking Save Changes

​

Saving Changes

All capability changes (standard and custom) are saved together in a batch operation:

1. Make all desired changes to your capabilities
2. Click Save Changes at the bottom of the page
3. A success message confirms the changes have been applied
4. Users will see updated capabilities immediately in their browser extensions

​

Best Practices

​

Agent Assignment

• Specialized Agents: Create agents specifically designed for each capability type
• Testing: Test agents in the platform before assigning them to browser extension capabilities
• Performance: Choose agents optimized for quick responses since users expect real-time interaction
• Chat History: Enable the “Include the chat history” option in AI steps for agents deployed to the browser extension. This maintains conversation context across interactions within the same browsing session.

​

Custom Capability Design

• Clear Names: Use action-oriented names that clearly describe what the capability does
• Focused Tasks: Design capabilities for specific, well-defined tasks
• Mode Selection: Choose the appropriate mode based on what data the agent needs:
  • Use Page for analyzing entire documents
  • Use Selection for focused analysis of specific content
  • Use Page Source only when HTML structure is needed

​

Initial Messages

• Be Specific: Provide clear instructions about what the agent should do
• Set Expectations: Include output format requirements (e.g., “in bullet points”, “as a table”)
• Context Awareness: Remember that page content is automatically included

​

Use Cases

​

Standard Capabilities

• Research Teams: Enable Summarize and Key Points for quick literature review
• International Teams: Enable Translate for multilingual content access
• Content Teams: Use all three for content analysis workflows

​

Custom Capabilities Examples

• Sales Teams: “Extract company information” - captures company names, sizes, and industries
• Legal Teams: “Find compliance mentions” - identifies regulatory and compliance references
• Support Teams: “Diagnose error messages” - analyzes error logs and suggests solutions
• Marketing Teams: “Analyze competitors” - identifies competitive positioning and messaging

​

Troubleshooting

​

Shadow AI Security Policies

Beyond Quick Actions, the browser extension can enforce security and compliance policies on employees’ use of third-party AI sites (ChatGPT, Claude, Gemini, and Perplexity). These policies are configured separately from capabilities, at Security → Browser Extension, and can be assigned to specific users or groups, or set as your organization’s default policy.

​

Enterprise Identity Enforcement

Requires that users be signed in with an approved corporate account before they can use ChatGPT, Claude, or Gemini in the browser — blocking anonymous sessions and personal accounts. Configuration:

• Enable/Disable: Turn enforcement on for this policy
• Approved Email Domains: The corporate email domains allowed to sign in (e.g., company.com). Enabling enforcement with no domains configured blocks everyone, so the editor warns you if the list is empty.
• Enforced Sites: Apply to all supported sites, or select individually from ChatGPT, Claude, and Gemini
• Fail Mode: Choose fail-closed (block when the user’s account status can’t be determined — recommended for stricter compliance) or fail-open (allow access if detection is inconclusive)
• Block Message: Custom text shown to blocked users (up to 256 characters)

​

Prompt Scan Gating

Inspects outbound prompts to ChatGPT, Claude, Gemini, and Perplexity, blocking messages that contain sensitive content before they leave the user’s browser. Built-in scanning rules ship for all four providers out of the box.

​

Traffic Monitoring & Data Capture

Enables visibility into AI service usage across your organization for compliance reporting. Configuration:

• Capture Request Body: Whether to capture the body of outbound requests to detected AI services
• Max Body Size: Maximum size (in bytes) of request bodies to capture

​

Domain Rules

Define category-based rules (e.g., block, warn, or allow) for specific domains, in addition to the built-in AI site coverage.

​

Enterprise Deployment

For organizations using Mobile Device Management (MDM) systems, the browser extension supports centralized, pre-locked configuration — connecting users automatically to the correct Airia region or custom instance, with connection settings locked so they can’t be changed. See the Browser Extension MDM Deployment Guide for the full configuration schema and step-by-step instructions for Chrome, Edge, and Firefox.