Transcript:

In today’s unboxing, we cover the first portion of the process. If the bot cannot handle certain use cases or determines that human intervention is required, those interactions can be routed back to a human agent.

In this session, we focus on what happens when a customer calls your contact center or business, how those calls are directed to Agentforce Voice, and how AI agents can answer those queries.

We begin with Salesforce Service Cloud Voice and Agentforce Voice already enabled. The first step is determining where to configure and enable Agentforce Voice.

Within Salesforce, there is a single place to set up the entire system: Setup. Everything on the left side represents configuration menu options and features that can be enabled within Salesforce.

For today’s session, we search for Agentforce Voice and follow the step-by-step setup instructions. The first step is to create a voice-enabled agent using Agent Builder.

We start creating an agent that can handle voice calls.
For this demo, the use case is a hybrid work policy, so the bot is designed to answer questions related to that policy.

We define the agent’s purpose, configure basic details, and assign a user.
Based on the description provided, Salesforce automatically generates related customer question topics.

Next, we provide a data source—a document containing the hybrid work policy stored in Salesforce Data Cloud.
Any documents related to the content the bot should reference can be used as knowledge sources for answering queries.

After completing these steps, we create the bot.

Once created, the bot builder interface displays the topics customers may ask about.
For each topic, the bot must know what action to take. Salesforce includes a predefined action to answer questions using the knowledge base, which we apply across all generated topics.

From a configuration perspective:

• Messaging is enabled for testing

• The hybrid policy document is included as a data source

• Language is set to English

• Tone is set to casual

• Event logs will capture activity once the bot is used

With configuration complete, we activate the bot and test it through the messaging interface.

We ask whether working from home every day is allowed.
After correcting a typo, the bot responds correctly, explaining that employees must work in the office a minimum of three days per week.

We then ask which specific days can be worked remotely, and the bot again responds accurately.
This confirms the bot is functioning as expected.

Next, we connect the bot to voice routing.

We configure call routing and escalation, starting with routing incoming phone calls to the AI agent.
This is done using an Omni-Channel flow in Salesforce.

Within the flow:

• Every incoming phone call is routed

• The routing record ID tracks the interaction

• The destination is set to the AI agent created earlier

• A fallback queue handles escalation if needed

After saving and activating the flow, we provision a phone number and associate it with the Omni-Channel routing flow.

This connects all components:

• Customer calls the phone number

• Salesforce routes the call through the Omni-Channel flow

• The call reaches the Agentforce Voice agent

During initial testing, the call does not route correctly.
Reviewing the configuration reveals the telephone channel was not connected to the bot.

After adding and enabling the telephone channel and reactivating the configuration, the system is tested again.

This time, the call succeeds.

The AI agent answers questions about the hybrid work policy, explaining:

• Employees must work in the office at least three days per week

• Standard in-office days are Monday, Wednesday, and Friday

• Fully remote work is not permitted without approval

The conversation confirms the full end-to-end voice experience is working.

That concludes today’s Agentforce Voice unboxing.
Thank you for joining the session.