How to Enable an Agentforce Chatbot on Salesforce Experience Cloud Site

Have you ever visited a customer portal and struggled to find the information you needed? Most users don’t want to search through multiple pages or submit support tickets for simple questions. They want immediate answers.

This is where Agentforce becomes valuable. By connecting an Agentforce Service Agent to an Experience Cloud site, organizations can offer conversational support directly within the portal using Embedded Messaging and web-based chat deployment. Instead of forcing users to navigate multiple pages, the chatbot can answer questions, guide users through actions, and route conversations when escalation is needed.

Experience Cloud portals are commonly used for support, partner collaboration, account management, and self-service experiences. Adding Agentforce introduces a natural-language support layer that helps users interact with the portal more efficiently.

Key benefits include:

Imagine an automobile service company that uses Experience Cloud as a customer self-service portal. Customers log in to book service appointments, check repair status, download invoices, review service history, and raise complaints.

Without a chatbot, support teams may receive the same questions every day, such as:

With Agentforce embedded in the Experience Cloud site, the company can answer these repetitive questions instantly, reduce support workload, and improve self-service adoption.

A typical implementation follows this flow:

This architecture works well because the Experience Cloud site provides the user interface, the messaging layer carries the conversation, and Agentforce handles the AI-driven support experience.

Before starting the implementation, confirm that the org is prepared correctly. Salesforce states that Einstein Generative AI and Agentforce should be enabled before launching agents, and the right access and permissions should be in place.Make sure the following prerequisites are covered:

The first step is to build the AI agent that will power the chatbot. Salesforce says agents should be created in Agentforce, configured with business instructions, tested, and activated before they are deployed to an external experience.

Navigate to:

Enable both features if they are not already active in the org.

Define the agent’s behavior using instructions, welcome text, fallback messaging, and business actions.A strong instruction example would tell the agent to help users check appointment status, view service history, raise support requests, and answer portal-related questions.Recommended configuration includes:

Use the preview or testing interface in Agentforce Studio to validate expected questions and responses before going live. After testing, activate the agent version because only active agents can be used in the deployment flow.

The next step is routing. In practical implementations, routing helps determine whether the conversation stays with the chatbot, triggers a flow, or gets escalated to a human support queue when needed.

Navigate to your flow and messaging routing setup area in Salesforce and create the logic for inbound conversations. This part is especially important if the business wants the chatbot to hand off complex or sensitive issues to human agents.

Now create the site where the chatbot will be exposed. Salesforce says the site must exist in Experience Cloud before the embedded messaging component can be added.

Navigate to:

Enable Digital Experiences if it is not already enabled, and configure the Salesforce domain if prompted during setup.

Navigate to:

Then:

After the site is created, open Experience Builder.

The messaging channel connects the Experience Cloud site to the Agentforce support experience. In Salesforce’s current messaging-based setup pattern, the messaging channel should be created before the web deployment that surfaces the chatbot on the site.

Navigate to the messaging setup area in Salesforce and create a new web messaging channel. Associate it with the Agentforce agent and the routing logic prepared earlier.

Suggested channel configuration includes:

The messaging channel acts as the bridge between the Experience Cloud site, the deployment layer, and the AI agent. Without it, the chatbot cannot communicate with portal users through the web interface.

Once the channel is ready, create the deployment that will actually be referenced inside Experience Builder. Salesforce says to create an embedded chat deployment for each chat window and connect it to the correct configuration.

Navigate to:

Then:

You should capture the following values for the Experience Builder step:

These settings are required because the Embedded Messaging component needs them to connect the site to the deployed chatbot experience.

At this stage, the chatbot is ready to be placed on the portal. Salesforce says to use the Embedded Messaging component in Experience Builder and place it in a visible site region, commonly the footer.

Navigate to: