Salesforce Agentforce Tutorials

The customer support team struggled to keep up with the increasing number of customer inquiries. Customers frequently ask about order status, software installation issues, and billing inquiries, most of which could be resolved without human interaction.

To avoid manual handling and human interaction for such tasks, Salesforce introduced an AI assistant tool, Agentforce, which is an AI assistant for Salesforce CRM.

In this tutorial, we will learn about the introduction to Salesforce Agentforce. We will see different agents and the building blocks of an agent working autonomously, and I will explain what Agentforce is, its features, types of agents, topics, instructions, actions, and the Atlas reasoning engine.

Introduction to Salesforce Agentforce

The Agentforce in Salesforce is an advanced AI-driven tool designed to create and manage tasks automatically, enhancing business operations across various domains such as sales, service, marketing, and commerce. By using Agentforce, organizations can automate tasks, improve customer engagement, and boost operational efficiency.

In simple words, AgentForce is the platform for Salesforce users, partners, admins, and developers to build, test, and deploy AI agents in their Salesforce org or external website. AgentForce is nothing but the Einstein Copilot, which is now known as an agent for Salesforce users.

Similarly, we can deploy other AI agents to the customer, community, or website to help answer users’ questions. This would be a replacement for the chatbots we used to deploy on the website or portal.

AgentForce responds to queries and requests in a language inspired by human speech and provides relevant responses based on company information. It was designed to help employees with regular business interactions and securely increase efficiency. It can help employees across a wide range of workflows and tasks on your desktop and mobile devices.

The agent runs in the user context, so whatever permission the user has, the agent also has similar permissions. To keep the data secure and private, Salesforce uses the Einstein trust layer.

Uses and Features of AgentForce in Salesforce

AgentForce has everything we need to make AI agents accurate and reliable for business. Now, let’s see how AgentForce is used in Salesforce.

Using AgentForce, we can summarize Salesforce records, such as opportunities, accounts, cases, and all standard and custom objects, by retrieving relevant data.
These agents work alongside human employees to enhance customer experience and streamline business operations.
We can also use AgnetForce to use pre-built email templates to interact with customers.
The agents are assistants who can autonomously perform tasks, make decisions, and interact with customers or the internal system.
The AI agents understand and process information, communicate naturally with the users, and take actions to perform specific tasks.
We can extend the standard agent actions or build new agents in Salesforce according to business needs so that they can solve tasks specific to the organization.
The Agentforce platform also allows partners to build and deploy agents and agent actions through App Exchange, which users can add to their org.

Types of Agents in Salesforce

Internal Salesforce Agent: This agent is deployed on Salesforce’s internal platform, which helps the user in their daily work and can be accessed from anywhere within Salesforce org. This is a kind of global action available on different apps, such as sales, service, or any of the tabs; this global action is accessible, so we can toggle it and ask different questions.
Community Site Salesforce Agent: This agent is deployed on the community site and helps answer questions and perform various jobs for end users without needing a service agent. It can work 24 hours a day on the website and handle multiple queries.

Topic, Instruction, and Actions in Salesforce Agentforce

Now, let’s understand the building blocks of an agent and the work behind the scenes of agents. These are things called topics, instructions, and actions.

Topic in Salesforce Agentforce

The topic defines the work these AI agents will do, so each topic defines the jobs the agent needs to perform. Here, we defined the job in natural language.

Unlike chatbots, where we define a decision tree, these topics contain a set of instructions and actions that will perform a task while creating a topic.

When creating a topic, we define the instructions the agent will follow if it chooses this topic for the current request.

Instructions in Salesforce Agentforce

Instructions are kind of rules that the agent should follow while solving customer queries. For example, we can give instructions that do not accept negative values, define a rule that the zip code length should be 5 digits, require the user to provide the order number, etc. Instructions are the guardrails that ensure that the Agent will perform the task within our defined boundaries.

For example, we need to define the required information in the instructions for asset installation, and creating an installation request is the actual action. However, we can define how to perform the action using instructions.

Actions in Salesforce Agentforce

Actions actually perform the task, like creating a draft order, booking an Appointment, fetching some Account details, etc. We will create standard and custom actions in the system according to the organization’s needs, which will solve a particular task.

The custom actions can be created using Flows, Apex, and prompt templates.

Standard Actions: When we enable the agent, the actions already provided by the Salesforce are standard actions.
Custom Action: Salesforce users create custom actions according to business needs to solve tasks specific to the organization.

Atlas Reasoning Engine in Salesforce Agentforce

Atlas Reasoning Engine in Salesforce Agentforce is designed to simulate human-like thought processes within the Salesforce Agentforce platform. Whenever the agent receives a query or a task to perform, the Atlas engine gets activated, and it is processed by AI so that it can do a lot of things on its own.

Atlas reasoning engine performs the following operations:

Identify the query intent.
Converse with the user.
After understanding the requirements, it prepares the plan to perform the activity, and it also stores the information in memory.

The Atlas engine is the brain of the agents, making decisions, asking questions, processing data, and performing actions.

Salesforce Agentforce Tutorials

Conclusion

I hope you have got an idea about Agentforce in Salesforce. We have explored various agents and their building blocks, which enable autonomous operation. I have explained what Agentforce is, its features, types of agents, topics, instructions, actions, and the Atlas reasoning engine.