Skip to main content

Website KA

Creating a Website Knowledge Agent

Step 1: Access the Agent Screen

  1. From the main navigation sidebar, click Agents .

  2. Click the blue + Create Agent button at the top left.

  3. You will see the Agent Type Library. Locate the Knowledge Agent card (usually under AutoMate Pack) and click it.

Step 2: Goal – Select Knowledge Agent

  • The first tab, Goal, is now complete. You have selected Knowledge Agent as your type.

  • You will automatically move to the Define tab.

Step 3: Define – Basic Setup

3.1. Agent Name

  • In the Agent name field, enter a descriptive name (e.g., "Website Support Assistant" or "Event Knowledge Bot").

3.2. Knowledge Agent Type (Channel Selection)

  • You will see channel options: Website, Email, WhatsApp, Telegram.

  • Select Website . This configures the agent for website deployment.

3.3. Where to Run This Agent (Targeting Rules)

Specify which pages on your website will have the agent active:

Option

Description

Full Domain

Agent appears on every page of the selected domain.

Section of Domain

Agent appears only in a specific section (e.g., /blog/).

Single Page

Agent appears on one specific URL.

Regex

Advanced URL pattern matching.

  • Choose your desired option and configure the URL accordingly.

  • If available, you can also filter by Start date, Categories, or Tags (depending on your CMS integration).

Note on Content Filters (Start Date, Categories, Tags):

The availability of Start date, Categories, and Tags filters depends on your domain configuration.
If these filters are not visible or grayed out, it means your domain has not been configured to sync this metadata from your CMS.
You must first complete the domain configuration process to enable these filtering options.

How to configure domain settings

  • If these filters are not visible or grayed out, it means your domain has not been configured to sync this metadata from your CMS.

  • You must first complete the domain configuration process to enable these filtering options.

3.4. Exceptions (Optional)

  • Add specific URLs where the agent should not appear, even if they match the rules above.

  • Enter the URL and click Add .

3.5. Select Language

  • Choose the primary language of your website (e.g., English).

3.6. Navigate to Design

  • Click Next to proceed to the Design tab.

Step 4: Design – Configuring the Agent Experience

This is where you attach your Engine, configure questions, CTAs, and design the look and feel.

4. Targeted Urls (Knowledge Base Selection)

Before configuring how your agent behaves, you must tell it what it should know.

  • Targeted Urls : This is where you choose the KA Collection that will serve as your agent's knowledge base.

  • Click the box and select the collection you created earlier (e.g., "KA for BE website," "Product Documentation").

  • This collection contains all the articles, PDFs, and web pages the AI will read to answer user questions.

  • you can attach Up to 3 collections to your agent

Important: Your Collection is the Source of Truth


The Collection provides the factual knowledge your agent relies on. Without a selected collection, your agent has no information to answer questions.

Ensure you have created a KA Collection in the Collections section before proceeding. If no collections appear in the sidebar, go back and complete the Collection Creation Guide first.

See the KA Collection Creation Guide

4.1. Settings and Configuration

  • Select engine: Click the dropdown and choose the KA Engine you created earlier. This attaches the "brain" to your agent.

Important: Selecting Your Engine

The Engine is the brain of your Knowledge Agent. It defines:

  • Tone of voice (formal, casual, friendly)

  • Answer structure (how responses are formatted)

  • Intent detection (categorizing user questions)

  • Fallback behavior (what happens when no answer is found)

Ensure you have created a KA Engine in AI Labs → Engines before proceeding. If no engines appear in the dropdown, go back and complete the Engine creation guide first.


See the KA Engine Creation Guide

Intent-based CTA Configuration

This advanced feature allows you to attach specific Calls to Action (CTAs) based on the user's detected intent during a conversation. It leverages the Intents you already defined in your KA Engine (during the Intent Guidelines setup) and triggers relevant CTAs when the AI identifies that intent.

How It Works

Intents are defined in the Engine: When you created your KA Engine, you configured Intent Guidelines and added specific intents The AI uses these to categorize what the user is asking about.

NOTE: IF you do not add intents on your engine, the "Intent-based CTA Configuration" will be hidden on the wizard

How to Configure Intent-based CTAs

Step-by-step

  1. Locate the Section:

    In the Design tab, scroll down to the Intent-based CTA Configuration section. It is placed before the Placement section.

    You will see a note: "Note: How intent-based CTAs work", click this for a quick refresher if needed.

  2. Select an Intent:

    Click the Select intent dropdown. This will display a list of all the intents you created in your linked KA Engine (e.g., pricing_query, partnership_inquiry, technical_support).

    Choose the intent you want to configure.

  3. Attach a CTA:

    After selecting an intent, click the Add CTA button that appears.

    A CTA selection modal will open, showing your existing CTAs with an option to Create new .

    Select the desired CTA (e.g., an email capture form for pricing_query, a demo booking link for partnership_inquiry).

  4. Repeat for Multiple Intents:

    You can add as many intent-to-CTA mappings as needed. Each intent can have its own dedicated CTA.

  5. Manage Your Mappings:

    Once added, each mapping will appear in the list, showing which intent is linked to which CTA. You can edit or delete mappings as needed.

  • Suggest Questions to visitors: Toggle this ON to allow the system to generate and display relevant questions related to the article. This enhances engagement by giving users starting points.

    • Question Type: If toggled on, you can select the type of questions (AI Generated or Static questions).

    • Add Questions: You can manually add custom questions that will be displayed to users.

      • Enter a question in the Static Questions field and click + Add .

  • Show CTA: Toggle ON to require users to complete a Call to Action before accessing Knowledge Agent responses.

  • Make the CTA mandatory: If enabled, the close button becomes invisible, users must complete the CTA.

  • Show CTA After Responses: Set the number of responses after which the CTA will appear.

  • Add CTAs: Click ADD CTA to select or create a new CTA (e.g., Email collection, Link, Dynamic Form).

  • Show Static CTAs: Toggle ON to show static CTAs within the agent responses.

  • Add End Screen: Click ADD End Screen to configure a final message after the conversation ends.

4.2. Placement (Where on the Page)

Configure Desktop Placement and Mobile Placement separately.

If you do not configure mobile placement, the bridged system will automatically pick the same placement for desktop and apply it on mobile

Insertion method: Select how the agent card is embedded:

Method

Description

Paragraph Number

Insert after a specific paragraph number.

After a Pattern

Insert after a specific word or phrase appears in the text.

HTML Element

Insert near a specific HTML element (e.g., by Element ID).

Side Panel

The agent appears as a collapsible side panel (common for chat widgets).

  • Auto Expansion: Toggle ON to have the card automatically expand when the user interacts with it.

  • Spotlight Animation: Toggle ON to draw attention to the agent when it loads.

4.3. Visualization (Look and Feel)

This section controls the visual appearance of your agent.

  • Theme Tab:

    • Template type: Select a template (e.g., Inline, side panel).

    • Pick a primary color: Choose your brand color.

    • Dark Theme: Enable for a dark mode version.

    • Adjust corner radius, margin, box shadow, and animated border as desired.

    • Configure Desktop and Mobile width settings (e.g., Max width in pixels).

  • Elements Tab:

  • Knowledge Agent Title: Set the title displayed in the agent.

  • Text Area Placeholder: The placeholder text in the input box (e.g., "Ask me anything...").

  • Submit Button Text: Customize the send button text. (This is useful only for few templates)

  • References Section Title: Title for the section showing source links.

  • Enable ratings: Allow users to rate responses.

  • Footer Settings: Customize the footer.

  • Bridged Watermark: Toggle the "Powered by Bridged" text on/off.

  • Fonts Tab:

    • Configure separately for Desktop and Mobile .

    • Choose Google Font or Custom Font .

    • Adjust font scale and select specific fonts.

  • Logo Tab:

  • Settings:

    • Upload your brand logo (JPG/JPEG/PNG, max 5MB).

    • Adjust Logo scale for desktop and mobile.

  • More Settings Tab:

    • Access advanced options like Z Index and Background Position .

4.4. Button Settings (For Side Panel Mode)

If you selected Side Panel as your placement method, configure the button that opens the panel:

  • Desktop Side Panel Button / Mobile Side Panel Button:

    • Style: Choose between text-only, icon-only, or both.

    • Text: Set the button text (e.g., "Ask AI").

    • Icon: Upload or select an icon, adjust size.

    • Button Position: Choose where the button appears (e.g., Bottom Right).

    • Background Color: Set the button color.

    • Gradient Animation: Enable animated gradients.

    • Floating Side Panel: Toggle to make the panel float with a margin on all sides.

    • Tooltip: Add a tooltip that appears on hover (with customizable text).

  • Content Tab:

    • Configure the main texts appear on the button

  • Interaction Tab:

    • Configure any interactive behaviors (if available).

Once all design elements are configured, click Next to proceed to Publish.

Step 5: Publish – Going Live

5.1. Smart Scheduling

  • Go Live Date: Set the date and time your agent should start engaging visitors.

  • End Date (Optional): Set an automatic deactivation date if needed.

5.2. Final Submission Options

You have two choices:

  • Submit without launching : Saves as a draft (visible in the Draft folder). The agent is not live.

  • Submit and launch : Makes the agent live immediately or at the scheduled Go Live Date.

Click your chosen button to complete the process.