We have a new tutorial on How to Build an Agentforce Voice Agent. Check it out here
Agentforce is all the buzz right now, and for good reason! This powerful tool is transforming how businesses approach customer service by integrating AI into their workflows. At AMC Technology, we recently hosted a quick workshop on building an Agentforce agent, and we couldn’t be more excited to share the insights from that session with you. Whether you’re looking to optimize your contact center or dive into the world of AI-powered agents, this step-by-step guide will show you how to create and implement an Agentforce agent that elevates your customer service operations.
Ready to see the workshop in action? Watch the full webinar here and dive deeper into Agentforce! Let’s get started and see why Agentforce is the solution everyone’s talking about!
Let’s get started!
What is Agentforce?
Before diving into the step-by-step guide, let’s first understand what Agentforce is and how it can help you build AI-powered agents.
Overview of Agentforce
Agentforce is an AI-driven platform designed to help businesses create and deploy virtual agents for customer support, lead generation, and workflow automation. It integrates seamlessly with Salesforce, allowing organizations to enhance their contact center operations by leveraging AI-powered responses, automation, and intelligent routing.
Key Features of Agentforce
- Pre-Built and Customizable Templates – Start with ready-made templates or build an agent from scratch.
- Natural Language Processing (NLP) – Enables agents to understand and process user queries effectively.
- Machine Learning (ML) Integration – Helps improve accuracy and responsiveness over time.
- Omni-Channel Support – Deploy agents across multiple channels, including chat, email, and voice.
- Case Management Automation – Streamlines issue resolution by automating case creation and routing.
Why Use Agentforce?
Building an agent with Agentforce allows businesses to improve efficiency, reduce response times, and provide customers with intelligent self-service options. Whether you need an agent for answering FAQs, handling customer support cases, or assisting with sales inquiries, Agentforce provides the tools to develop and deploy AI-powered assistants quickly.
Step 1: Register on Trailhead
- Visit Trailhead Registration.
- Sign up for a free account if you don’t have one already.
- Complete the initial setup steps as prompted
- Keep in mind that these orgs expire in six months.
Step 2: Turn On Einstein and AgentForce in Setup
Once your Salesforce org is set up, you’ll need to configure Einstein and AgentForce.
- Search for “Einstein” in Quick Setup.
- In Quick Setup, type “Einstein” and make sure the “Turn on Einstein” toggle is enabled.
Search for “AgentForce” in Quick Setup.
- Type “AgentForce” into Quick Setup and ensure the toggle is also turned on.
Step 3: Create a New Agent
Click on the “+” to create a new agent.
- Select the agent type as “AgentForce Service Agent.”
Name the agent.
- For example, “Oscar Party Service Agent” and add “demo” in parentheses.
Provide a description.
- Add a description explaining what the agent does (this can be generic or specific to your demo)
Define Roles, Permissions, and User Settings
To ensure the agent can perform its tasks, define its role and permissions.
Assign a Role.
- Provide the role for the agent (e.g., answering Oscar-related questions and handling customer service tasks like processing returns).
Assign Permissions.
- Select a user with the appropriate permission sets that allow them to use AgentForce and Service agents.
Step 4: Create Topics
Now, you need to define the topics your agent will handle.
Create a New Topic (e.g., Q&A).
- Click on “+ New Topic” and name it “Q&A.”
- Provide a description and scope, explaining that it handles frequently asked questions.
Set Instructions for the Topic.
- Add instructions for the agent to follow, such as “When the user asks a question related to Hollywood, run the action ‘Get FAQ’ and return the results to the user.”
Create Additional Topics (e.g., Profile Management).
- Repeat the process to create topics like “Profile Management.”
- Provide relevant scope and instructions, such as returning profile information when requested.
Step 5: Configure Case Management
For agents handling customer service requests, you’ll need to configure a case management topic.
Adjust the Case Management Topic.
- Create a new version of the case management topic and add instructions.
- Example: “Verify the user’s identity before creating a case.”
Add Relevant Actions.
- Select actions like “Create Case” or “Get Contact.”
- These actions are predefined by Salesforce but can be customized based on your use case.
AgentForce Actions: Prebuilt vs. Custom
Understanding Actions in AgentForce
Actions define what an agent can do when responding to a user query. There are two types of actions:
Prebuilt Actions (Out-of-the-Box by Salesforce)
These include:
Create Case – Automatically creates a case for a user request.
Get Contact – Retrieves customer information from Salesforce records.
Get FAQ – Fetches common Q&A responses.
Best Used For: Standard workflows where customization isn’t required.
Custom Actions (User-Defined)
You can create your own actions using Apex code, Flow actions, or API calls.
Examples:
Send SMS Notification – Integrate Twilio or another SMS provider.
Check Order Status – Pull live data from an external order management system.
Best Used For: When prebuilt actions don’t meet your needs, and you require tailored solutions.
Step 6: Test the Agent’s Topics
Now that the agent and topics are set up, you can test them to ensure they’re functioning correctly.
Test the Q&A Topic.
- Ask questions like, “Can anyone visit the Hollywood Sign?”
- The agent should provide relevant FAQs or information.
Test the Case Management Topic.
- Simulate a case scenario, such as creating a case for ordering the wrong decoration items.
- Ensure the agent asks for necessary information (like email) and creates a case.
Test Profile Management.
- Ask the agent to retrieve profile information (email, phone number, etc.).
- Ensure it returns the correct profile details.
Step 8: Enable Embedded Service Deployment
To deploy the agent as an embedded service on a website or web application, you need to configure Embedded Service Deployment.
Navigate to Embedded Service Settings.
- In Salesforce Setup, search for “Embedded Service Settings.”
Create a New Embedded Service Deployment.
- Click on “New Deployment” and give it a name (e.g., “Oscar Party Service”).
Configure Embedded Service Settings.
- Choose the necessary options, such as channel type (chat or messaging) and appearance (color scheme, branding).
Deploy Embedded Service Code to Your Site.
- Salesforce will provide a snippet of code to embed the agent into your website.
- Copy this code and insert it into the HTML of your website or web application.
Step 9: Create a Flow for Interaction
Flows allow the agent to interact with customers in a more guided and customizable way. To build a flow, follow these steps:
Create a New Flow in Salesforce.
- In Setup, search for “Flow” and select “Flows” under Process Automation.
- Click “New Flow” and select a flow type (e.g., Screen Flow).
- Image Suggestion: Screenshot of the Flow setup screen.
Add Flow Elements.
- Use elements like “Screen” for displaying prompts, “Get Records” for fetching data, and “Create Records” for generating cases or leads.
- Example: Add a “Screen” element asking for the user’s issue and “Get Records” to retrieve account information.
- Image Suggestion: Screenshot showing the elements in the Flow builder.
Configure Flow to Interact with the Agent.
- Link the flow to the agent by using flow actions within the embedded service.
- Ensure the flow is triggered by specific topics or queries, such as when a customer asks about a product or service.
- Image Suggestion: Screenshot of the flow connected to the agent’s topics and actions.
Activate the Flow and Test.
- Once the flow is set up, activate it, and test the interaction between the flow and the agent within the embedded service on your site.
- Image Suggestion: Screenshot of the flow in action on the embedded service widget.
Step 10: Final Testing and Troubleshooting
Test the agent one last time to ensure it works seamlessly with all topics and flows.
Test all created topics (Q&A, Case Management, Profile Management).
- Verify each topic is working as expected by asking questions and interacting with the agent.
Test Flow interactions.
- Test if the flow properly guides the user through the questions and creates records if needed.
Troubleshoot any issues.
- If any topics or flows aren’t working as expected, review the instructions and actions to ensure they’re properly configured.
DaVinci Integration with Salesforce and Agentforce
DaVinci for Agentforce enhances your AI-powered agents by seamlessly integrating with Salesforce, transforming your contact center operations. By automating key processes and enabling smarter workflows, it allows your team to deliver personalized, real-time support.
- Caller Authentication: Automate identity verification during calls, improving security and efficiency.
- Automated Processes: Trigger actions based on real-time events, streamlining agent workflows.
- Personalized Service: Deliver tailored customer experiences through data-driven workflows across multiple systems.
With DaVinci, your Agentforce agents become more powerful, offering real-time transcription, agent assist features, and intelligent virtual assistants, all while reducing agent workload and enhancing customer satisfaction.
Visit our DaVinci for Agentforce to learn more about how we can streamline your customer service operations and improve your workflow efficiency.
FAQ: Building an AI Agent with Agentforce
Agentforce is an AI-powered platform designed to help users create and deploy intelligent virtual agents for customer service, lead generation, and workflow automation.
To build an agent with Agentforce, you need:
- A Trailhead account (sign up for free if you don’t have one).
- Access to the Agentforce platform within Trailhead.
- A clear purpose for your agent (e.g., customer support, lead generation).
- Sample data to train your agent’s responses.
No, Agentforce is designed to be user-friendly, so you don’t need coding skills. You can use pre-built templates and configuration options to create your agent.
Yes, you can deploy your agent across multiple channels, including web chat, email, and more, to provide omnichannel support.
Agentforce includes Natural Language Processing (NLP), machine learning-based improvements, integration capabilities, and automated response handling.
