Get Started with Einstein Agents and Copilot

Learning Objectives

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

Explain what Agentforce is.
Identify the main characteristics of Agentforce.
List a few use cases for Agentforce.
Describe how Agentforce helps you build a trusted AI agent.

Before You Start

We know you’re eager to get started, but we recommend that you consider completing these modules first so that you’re familiar with terms like LLMs, prompts, grounding, and more.

The Generative AI Revolution

We’re fast approaching the year 2025, and the 21st century hasn’t quite lived up to all of our futuristic expectations: We’re not jaunting off to vacations on Moon Base Alpha or zipping around in flying cars.

But even though we can’t journey to other worlds, we do live in an era where we can have complex, human-like conversations with artificially intelligent machines.

Not long ago, a new class of incredibly powerful AI models captured the world’s imagination with their ability to perform language tasks. To many it seemed like elements of science fiction had finally become reality. And the technology has sparked a lot of excitement about the incredible gains in productivity that generative AI will make possible.

The Rise of AI Assistants

Generative AI tools have quickly advanced by leaps and bounds, and we’re already seeing how the technology will change the future of work. One such advancement is the rise of AI assistants, also referred to as AI agents.

The latest breed of AI agents are intelligent trusted advisers that harness the power of large language models (LLMs) to communicate with humans using natural language. They’re designed to work alongside users, and they can boost productivity by automating mundane tasks, analyzing data, answering questions, providing contextual assistance, and more.

Sounds pretty great, right? But how do you take advantage of this technology and deliver on the promise of increased productivity? How do you build safe AI agents and integrate them into the flow of your users’ work?

Meet Agentforce

We’d like to introduce you to Agentforce, a new and trusted conversational AI agent for CRM that’s seamlessly built into the Salesforce interface. It handles questions and requests posed in natural language and provides relevant answers drawn from secure, proprietary company data. It’s built from the ground up to boost efficiency in a safe way by assisting employees with everyday business interactions. And it’s capable of helping employees across a wide range of workflows and tasks on your desktop and mobile devices.

In the conversational UI, a Salesforce desktop user asks Einstein to summarize a deal, and the assistant returns an opportunity summary.

Best of all, you don’t need to know a stitch of code to set up an assistant in Salesforce. All you have to do is enable Agentforce, and right out of the box your assistant can assist your users with common tasks in Salesforce. Your assistant can:

Summarize Salesforce records, such as opportunities, accounts, and cases.
Draft or revise sales emails.
Find Salesforce records.
Aggregate Salesforce data.
Answer questions with information from your knowledge base.
And more!

On top of that, it’s easy to extend your assistant by using your existing Salesforce Platform functionality. For example, if you already have a flow in Salesforce that can make product recommendations, you can add that capability to your assistant with just a few clicks.

Characteristics of Agentforce

Agentforce is a trusted, natural language, conversational AI agent. Let’s take a closer look at each of those characteristics so you can get more familiar with your new digital companion.

Trusted

Agentforce can take action in your Salesforce org based on a user’s request. The assistant sends the user’s request to the LLM, and the LLM generates and executes a plan to carry out the request.

Your assistant respects standard Salesforce access controls like licenses and permissions, so it will always act securely. Agentforce is also integrated with the Einstein Trust Layer, which is a secure AI architecture natively built into the Salesforce Platform.

Designed for enterprise security standards, the Trust Layer allows you to benefit from generative AI without compromising your customer data. At the same time, it lets you use trusted data to improve generative AI responses.

Data grounding: The Trust Layer grounds and enriches generative prompts in trusted company data.
Zero-data retention: Your data will never be retained by a third-party LLM provider.
AI monitoring: AI interactions are captured in event logs, giving you visibility into the results of each user interaction.

At Salesforce, Trust is our #1 value. That’s why, as we build generative AI tools like Agentforce, we carry our value of Trust right along with us.

Natural Language

Agentforce is a conversational interface, so employees can express their questions or instructions in natural language as if they were talking to a human.

For example, instead of clicking around the Salesforce UI to find records, a sales rep can launch Agentforce and simply type, “Show me my Acme deals.” The assistant, powered by an LLM, interprets the request, responds to the user in natural language, and displays a list of matching opportunities. The interaction feels as familiar as having a conversation with a trusted colleague.

In the conversational UI, a Salesforce mobile user asks Einstein for their top opportunity, and the assistant shows the opportunity.

The Einstein panel, which is a conversational interface, is currently available in Lightning Experience and Salesforce Mobile.

And speaking of natural interactions, your mobile employees can use voice commands to ask questions or give instructions to Agentforce. This makes business tasks much easier to accomplish for users on the go.

Conversational

Finally, Agentforce is a conversational agent, so each user request or instruction is understood in the context of an ongoing dialogue. Some AI assistants are only capable of having a single back-and-forth interaction with a user. But with Agentforce, a user can ask a follow-up question or make a related request, and the context is retained.

In the previous example, a sales rep asked to see their Acme deals. In the same conversation, they could then ask, “Show me their open cases.” Agentforce remembers the conversation history and knows that the word “their” refers to Acme, so it’s capable of having multiple back-and-forth interactions with a user.

Agentforce Versus Einstein Bots

Not sure about the difference between Agentforce and Einstein Bots? Bots require a lot of expertise and time to set up, and they’re based on complex, strictly defined conversational rules.

Agentforce is more flexible and requires less configuration because it’s powered by an LLM. For more information, see the blog post in the Resources section.

The Hero of Our Journey

Excited about the possibilities? Ready to learn more about Agentforce? Awesome. In this module, we explore Agentforce in the context of a real-world scenario. Let’s meet Linda Rosenberg, the Salesforce admin at Cloud Kicks.

Portrait of Linda Rosenberg.

Cloud Kicks sells stylish and comfortable custom sneakers, and the company is growing fast. Linda is always looking for ways to make her end users’ work easier, and she’s excited about the value that an AI agent like Agentforce can provide. Her goal is to increase employee productivity by automating time-consuming and repetitive business tasks, which will free people up for more strategic activities.

In this module, you learn how Linda customizes, tests, activates, and monitors an assistant in Salesforce. So without further ado, let’s tackle the next unit, where Linda explores the different components of Agentforce and learns more about how the feature works.

Looking for Help?

Discover More