Salesforce Einstein Copilot, with its robust standard actions and customizable capabilities, is transforming the way businesses interact with their CRM. Join us to explore Einstein Copilot Standard and Custom Actions.

Introduction to Einstein Copilot

Salesforce Einstein Copilot is a groundbreaking AI-powered assistant designed to seamlessly integrate into the Salesforce interface, enabling users to perform tasks through conversational interactions. This AI-driven tool is more than just a chatbot; it’s a powerful assistant that can execute business tasks, streamline workflows, and provide intelligent insights. In this post, we’ll dive deep into the two types of actions that drive Einstein Copilot—standard and custom actions—and how they can be leveraged to enhance your Salesforce experience.

How Einstein Copilot Works

Before we explore the actions, let’s understand the basic components of Einstein Copilot:

• Copilot: The AI assistant that interacts with users through conversational language.
• Actions: The tasks that the copilot can execute on behalf of users.
• Reasoning Engine: The intelligent system that interprets user requests, plans actions, and executes them to achieve the desired outcome.

Standard Actions in Einstein Copilot

Standard actions come pre-built within Einstein Copilot, allowing it to perform a variety of common Salesforce tasks right out of the box. These actions are integral to the basic functionality of the copilot and are essential for everyday business operations.

Key Standard Actions Include

• 3.1 Identify Record by Name (System Action):
  • Functionality: Searches and retrieves Salesforce records based on a specified name, returning a list of matching record IDs.
  • Example: A sales manager asks, “Find the account records for ‘Tech Innovations,'” and Einstein Copilot quickly identifies and presents the relevant account records for review.
• 3.2 Identify Object by Name (System Action):
  • Functionality: Determines the specific Salesforce object referenced by the user, enabling further actions on that object.
  • Example: A user types, “Show me all open cases for Tech Innovations,” and the copilot identifies that the request relates to both the Account and Case objects, guiding the user to the relevant case records.
• 3.3 Query Records (Beta):
  • Functionality: Retrieves Salesforce records based on user-defined conditions, such as filtering by field values and sorting results.
  • Example: A sales analyst queries, “List all opportunities closing next month with a value over $50,000,” and Einstein Copilot returns a filtered list of opportunities that meet these criteria.
• 3.4 Query Records with Aggregate (Beta):
  • Functionality: Aggregates Salesforce data to provide answers based on operations like count, sum, average, etc.
  • Example: A marketing director asks, “What’s the average deal size for opportunities closed this quarter?” and the copilot calculates and presents the average deal value.
• 3.5 Summarize Record:
  • Functionality: Generates a concise summary of a Salesforce CRM record, highlighting key details.
  • Example: An account executive needs a quick overview of a complex deal and asks, “Summarize the key points of the Acme Corporation opportunity,” resulting in a brief summary that highlights the deal stage, key contacts, and next steps.
• 3.6 Draft or Revise Sales Email:
  • Functionality: Automatically composes or revises a sales email, using Salesforce data to ensure personalized communication.
  • Example: A salesperson asks, “Draft a follow-up email for the lead ‘Jane Doe’ regarding our recent product demo,” and the copilot generates an email draft that references specific points discussed during the demo.
• 3.7 Answer Questions with Knowledge:
  • Functionality: Provides answers based on information from Salesforce knowledge articles, ensuring accurate and reliable responses.
  • Example: A customer service representative asks, “What’s our policy for processing returns over 30 days?” and the copilot retrieves the relevant return policy from the knowledge base and presents it clearly.

These standard actions empower Einstein Copilot to handle a wide range of common tasks, making it an invaluable tool for Salesforce users.

Custom Actions in Einstein Copilot

While standard actions are powerful, custom actions take Einstein Copilot’s capabilities to the next level by enabling businesses to tailor the copilot to their unique needs. Custom actions are built on top of existing Salesforce functionalities, such as invocable Apex classes, autolaunched flows, and prompt templates.

Creating Custom Actions

1. Identify the Process: Determine the specific business process or workflow that needs automation.
2. Leverage Existing Functionality: Use Salesforce Flow or Apex to create the action that will be invoked by the copilot.
3. Integrate with Copilot: Configure the custom action in the Copilot Builder and assign it to the appropriate copilot.
4. Test and Refine: Ensure the custom action works as intended by testing it with real data.

Examples of Custom Actions

• Get Detailed Order Status: A custom action that retrieves and displays detailed order status, including shipping information, expected delivery dates, and any potential delays, based on a customer’s order number.
  • Use Case: A support agent types, “What’s the status of order 12345?” and Einstein Copilot pulls up comprehensive details, helping the agent provide accurate information to the customer.
• Initiate an Order Return: Automate the process of initiating a return for a customer order, including generating return labels and updating the order status.
  • Use Case: A customer service rep asks, “Start a return for order 12345,” and the copilot kicks off the return process, sending the necessary information to the customer and updating the system.
• Schedule Sales Meetings: Create a custom action that coordinates with the sales team’s calendars to schedule meetings with leads or customers directly from a conversation.
  • Use Case: A sales manager requests, “Schedule a meeting with John Smith next Tuesday afternoon,” and the copilot finds an available slot, sends an invite, and updates the Salesforce record.
• Generate Customized Proposals: Use Salesforce data to create a tailored sales proposal document that can be sent directly to the client.
  • Use Case: A salesperson says, “Generate a proposal for Acme Corp with our premium package,” and the copilot assembles a proposal document including all relevant details, ready for review and dispatch.

Key Prerequisites for Custom Copilot Action

1. Verify that Einstein setup is activated in your Salesforce organization. If it’s not available, access a Trailhead playground to explore and test the functionality.
2. In Quick Find box, search for “Einstein Settings”. Enable Einstein by toggling the switch.

3. In Quick Find box, search for “Einstein Copilots”. Enable Einstein Copilot for Salesforce by toggling the switch

4. Design and activate a flow named “Active Account”. Refer to the provided screenshot for an example of the activated account flow.

Create a Custom Copilot Action

Creating a custom Copilot action in Salesforce Einstein Copilot allows you to tailor the AI assistant to meet the specific needs of your business. Below is a step-by-step guide to creating a custom Copilot action, using an existing flow as an example.

Step-by-Step Guide to Creating a Custom Copilot Action

1. Access Einstein Copilot Actions:
   • In the Salesforce Setup menu, go to the Quick Find box and type Copilot.
   • Click on Einstein Copilot Actions from the results.
2. Initiate a New Copilot Action:
   • Click on New Copilot Action to begin the process of creating your custom action.
3. Reference an Existing Flow:
   • For Reference Action Type, select Flow from the dropdown menu.
   • In the Reference Action dropdown, choose the flow you want to reference.
4. For this example, we’ll create and select Active Account, a flow designed to automate the process of creating activating an account in Salesforce.

5. Define Action Instructions:
   • Copilot Action Instructions: Provide clear and concise instructions that will guide the AI in executing the action. For instance: ◦ “Use Activate Account for requests to activate an account”
   • Inputs Instructions: Specify what input data is required for the action. For example: ◦ “Populate this input variable with an Account Id” ◦ Check Require Input to ensure the necessary data is provided. ◦ Check Collect data from user if the data should be gathered interactively during the conversation.
   • Outputs Instructions: Describe the expected output from the action. For instance: ◦ “This output indicates the activation status of the request, whether it has been activated or not.”
   • Check Show in conversation to display the output during the conversation.
6. Complete the Custom Action:
   • After filling out the necessary details, click Finish to save your custom Copilot action.

7. The newly created custom Copilot action is now visible in the actions list.

8. Integrate the Custom Action into Your Copilot:
   • Navigate back to the Einstein Copilot section by clicking on Einstein Copilots in the setup menu.
   • Click on your existing Copilot and then click Open in Builder. • Before adding the new action, you’ll need to deactivate the Copilot temporarily. Click Deactivate, then click OK to confirm.
   • Go to the Copilot Action Library.
   • Find your new action, Activate Account, in the action list.
   • In the action menu, choose Assign to Copilot to add it to your Copilot’s capabilities.
9. Reactivate the Copilot:
   • Once you’ve added the new custom action, click Activate to make the Copilot available again with its newly enhanced capabilities.

10. Quickly test the Einstein Copilot after activating the copilot.
11. Navitage to Accounts, find einstein in the top and click on it to open the copilot window and provide your input.

Let’s check if the account record got updated.

Testing and Refining Your Custom Copilot Action

After creating and integrating your custom action, it’s crucial to test it thoroughly to ensure it performs as expected:

• Run Test Scenarios: Simulate user requests that would trigger the custom action to verify it works smoothly within the conversation flow.
• Collect Feedback: If possible, have a small group of users test the Copilot and provide feedback on the custom action’s performance.
• Refine Based on Results: Make any necessary adjustments to the flow, instructions, or action configuration based on the test results and user feedback.

Creating custom Copilot actions is a powerful way to extend the functionality of Einstein Copilot, allowing you to automate processes that are unique to your business. By following these steps, you can enhance your Salesforce environment, making it more responsive to your organization’s specific needs

How the Reasoning Engine Works

The reasoning engine, also known as the planner service, is the intelligence behind Einstein Copilot. It:

1. Interprets User Requests: Understands the intent behind user input.
2. Builds a Dynamic Plan: Creates a sequence of actions to achieve the user’s goal.
3. Executes Actions: Launches the appropriate actions based on the plan.

This dynamic and responsive approach ensures that Einstein Copilot not only understands what the user wants but also delivers the desired results efficiently.

Conclusion

Salesforce Einstein Copilot, with its robust standard actions and customizable capabilities, is transforming the way businesses interact with their CRM. By understanding and leveraging both standard and custom actions, organizations can optimize workflows, enhance productivity, and create a more intuitive user experience. Whether you’re automating routine tasks or developing complex workflows, Einstein Copilot is a powerful tool that can drive significant value for your business.