Learn About the Deployment Process

Learning Objectives

After completing this unit, you’ll be able to:

• Describe the deployment stage of the agent development lifecycle.
  
• Explain the components of a deployment strategy.
  
• Identify the available channels and connections for an Agentforce Service Agent.
  

In this badge, we assume you are a Service Cloud and Agentforce administrator with Omni-Channel enabled and the proper permissions to deploy Agentforce for Service Agents. If you’re not an administrator for Service Cloud and Agentforce, that’s OK. Read along to learn how your administrator would take the steps in a production org. Don't try to follow these steps in your Trailhead Playground. Agentforce Voice isn't available in the Trailhead Playground.

Before You Start

Before you start this badge, consider completing this recommended content.

• Agentforce Basics
  

Meet the Team Behind the Deployment

Northern Trail Outfitters (NTO) sells gear and clothing to outdoor enthusiasts. Its business is booming, and the support team is feeling the strain. Customers want fast answers about orders, returns, and product recommendations at all hours as the queue keeps growing.

So NTO built an Agentforce Service Agent. The team designed its subagents, wrote its instructions, built its actions, and tested the agent until it handled typical customer questions with confidence. The agent works beautifully in the organization’s sandbox environment, and it’s finally ready for use.

That's where Warren Mehta, an IT systems specialist at NTO, comes in. Warren's job is to take this tested agent and connect it to the channels where customers actually show up: the website, the phone line, and beyond. Let’s follow along with Warren throughout this badge as he deploys NTO's first AI service agent.

The Deployment Stage

Every Agentforce agent moves through the agent development lifecycle.

• Plan: Define the goals and audience for the agent, and align the agent with business needs.
  
• Build: Configure the agent's subagents, actions, and instructions in Agentforce Builder.
  
• Test: Evaluate the agent against real scenarios and refine its behavior.
  
• Deploy: Connect the agent to the channels where users interact with it.
  
• Observe: Monitor and track performance, and improve the agent over time.
  

NTO has finished planning, building, and testing. This badge focuses on the deployment of NTO’s agent.

The Deployment Process

So what does deploying actually mean? A built agent lives inside Salesforce, but your customers don't. Deployment is the work of connecting your agent to a channel (like a website chat window, a phone line, or a help center) so the people it's meant to serve can start a conversation. Deployment is the bridge between a great agent and a great customer experience.

Here’s the process in brief.

1. Create your deployment strategy.
   
2. Configure the agent’s connections in your sandbox environment.
   
3. Complete end-to-end testing.
   
4. Deploy the agent to your staging or production environment.
   

This badge explores each step of the process in more detail.

Build Your Deployment Strategy

Warren knows better than to rush straight into configuring the agent’s connections. A service agent talks directly to customers, so a rushed launch risks real consequences and damage. Before any configuration begins, the leadership at NTO makes sure that the implementation team has a deployment strategy for the AI agent.

A deployment strategy is your roadmap for maximizing the chances of your AI agent’s success and minimizing disruption. It helps you anticipate challenges, manage stakeholder expectations, and gather critical feedback for continuous improvement.

It’s a good idea to create your deployment strategy as you’re planning and prototyping your agent. Proper planning helps ensure that your AI agent rollout is a seamless and successful effort that delivers real value to your organization. Let’s go over the components of an effective deployment strategy.

Rollout Plan

The first step is to think about the logistics of the deployment. What’s your rollout plan for the AI agent? With an AI project, it’s always wise to start with a pilot program, which is a small, controlled group of users. The group might be a specific team or department in your organization. This allows you to test the agent in a real-world scenario without risking a companywide disruption. After refining the agent’s performance, you can gradually roll it out to other users.

Change Management and Governance

AI projects require robust change management strategies to reduce risk and ensure successful adoption. Here are a few things that can affect how quickly and successfully your organization and customers adapt to the new AI agent.

• AI maturity and organizational readiness
  
• Process redesign and workflow integration for AI tools
  
• Ethical AI governance
  
• Comprehensive training plan, resources, and documentation
  
• Effective communication strategy
  
• Executive sponsorship and user evangelism
  
• Culture of AI experimentation, continuous learning, and transparency
  
• Feedback and collaboration channels
  
• Incentive and recognition programs for AI adoption
  

For more information about change management strategies for AI projects, see the Resources section.

Connections and Channels in Agentforce

A channel is a messaging platform, app, or interface where your agent meets users: a web chat, a phone line, and more. When you deploy your agent to a channel, you create a connection. A connection bundles everything that ties the agent to that channel.

• Adaptive response formats (like images, buttons, and links) suited to that channel
  
• Omni-Channel flows that route conversations to and from the agent
  
• Context about how to reason and respond appropriately for one or more channels
  

One agent can have multiple connections. Because the agent itself stays the same, Warren can reuse one agent across a website, a phone line, and other channels, each with its own connection. He can configure the experience per channel without rebuilding the agent, which helps the NTO team scale agent development and reduce repetitive setup.

Where to Deploy

Now that you know more about connections in Agentforce, how do you start deploying your agent to different channels?

Let’s go over the deployment options for two of the most popular types of agents: Agentforce Service Agent and Agentforce Employee Agent.

Here are the available connections for these agents.

For the first iteration of its service agent, NTO will deploy to telephony and Enhanced Chat v1 channels.

Configure the Deployment in a Sandbox Environment

After building a deployment strategy and planning all the connections for the NTO AI agent, Warren is ready to start setting up the channels in NTO’s sandbox environment. In the next unit, you set up the messaging connection and route conversations to the agent.

Resources

• Salesforce Help: Deploy Your Agent to Channels
• Salesforce Help: Agentforce Implementation Guides
• Trailhead: Quick Start: Assemble a Service Agent with Agentforce Builder
• Trailhead: Change Management for AI Implementation
• External Site: AI Transformation: A People-Centric Guide to Leading Change
• Salesforce: Get Hands on with Agentforce Service Agents