What Are Agents?
Agents are goal-oriented, autonomous AI assistants that perform tasks and business and make data-driven decisions. Learn more about agents and how they compare to Einstein Bots.
Required Editions
| Available in: Lightning Experience |
| Available in: Enterprise, Performance, Unlimited, and Developer Editions. Required add-on licenses vary by agent type. |
Agents are goal-oriented, autonomous AI assistants that perform tasks and business interactions. They can initiate and complete a sequence of tasks, handle natural language conversations, and securely provide relevant answers drawn from business data. Some agent types are best at assisting and collaborating with a Salesforce user in the flow of work. Other types can also act on behalf of a user or customer, based on the use cases and guardrails an admin specifies. Agents adapt and improve with each interaction.
Some agent types are available in your Salesforce interface, and you can add some to your customer channels.
Agents respect standard Salesforce access controls such as licenses and permissions, so they always act securely. Agents also include AI guardrails designed to help meet ethical standards and are integrated with the Einstein Trust Layer, a secure AI architecture natively built into the Salesforce Platform.
Agents help with a variety of use cases.
| Use Case | How Agents Can Help |
|---|---|
| Intelligently automate repetitive tasks. | Agents perform common tasks autonomously, such as updating a Salesforce record, answering a simple question, or drafting an email. During a conversation, an agent determines which actions are required for the task at hand and performs those actions automatically. |
| Provide personalized customer service experiences. | Agents can generate context-aware responses to employee and customer inquiries. Their answers are securely grounded in business data. For more complex issues, agents can automatically escalate the conversation to a live agent. |
| Get data-driven insights about a specific Salesforce record or page. | Agents can assist users in their flow of work in Salesforce. For example, they can summarize a case and suggest actions based on the page a user’s viewing. |
| Get assistance with complex, creative tasks. | Agents can generate drafts for a variety of content types, including emails, newsletters, and text messages. They can also help users brainstorm and plan creative projects. |
Agentforce is included as a Covered Service in the Einstein Platform and Agentforce SOC 2 and SOC 3 reports. Agentforce is HIPAA eligible and covered under the Salesforce Business Associate Addendum Restrictions, and has also achieved ISO 27001, 27017, and 27018 certifications.
Differences Between Agents and Einstein Bots
Einstein Bots use predefined rules and scripted responses to navigate conversations. They excel at following strict processes and adhering to brand messaging guidelines.
When choosing whether to build an agent or a bot, consider your goals, your users, and the channels you want to deploy your conversational AI solution to. Each solution has different strengths.
Einstein Bots are well suited for scenarios where you want more transparency and control over conversations. Example scenarios:
- You’re in a regulated industry that requires explainable processes with extra governance.
- You want the chatbot to follow highly specific, deterministic conversation flows.
- You want to use generative AI in specific conversation flows only.
Some key differences between Einstein Bots and agents include:
- How they converse: Einstein bots offer scripted replies to user queries, making them suitable for straightforward or transactional interactions. Agents engage in fluid conversations and use conversation context to inform their decisions.
- How they process intent: Einstein bots are limited to predefined dialogs and rules. Because agents are built on LLMs, they’re more effective at interpreting the underlying intent of messages and providing accurate and contextually relevant answers.
- How they’re built: Einstein bots are built on simple intent models and follow hard-coded dialogs. They often require more extensive setup and training than agents. Agents are built on LLMs trained on vast amounts of data and decide what to do using subagents and actions.
| Feature | Agents | Einstein Bots |
|---|---|---|
| Built into the Salesforce interface |  Depends on the agent type |
|
| Can be added to enhanced messaging channels | Depends on the agent type |
|
| Integrated with the Einstein Trust Layer |
|
|
| Context of an ongoing conversation |
|
|
| Can leverage a knowledge base |
|
|
- The Building Blocks of Agents
Get to know the key components of agents, including subagents and actions, data, connections and channels, and the reasoning engine. - How Agents Work
Follow the path of a user message through the Atlas reasoning engine to learn how Agentforce agents reason, take action, and generate responses. - Agentforce Language Support
Learn about supported languages, multi-language setup, and agent testing with Agentforce Builder. - How Agents Created in the Legacy Builder Work
Follow the path of a user message through the Atlas reasoning engine to learn how an agent built in the legacy Agentforce Builder reasons, takes action, and generates responses. - Trust and Agentforce
With Salesforce, it’s easy to build safe AI agents because agents are designed to help mitigate the risks associated with generative AI technology. - Agentforce Glossary of Terms
Learn more about the terms used in Agentforce and generative AI at Salesforce. - Agentforce Considerations
Carefully review the support, limits, and considerations for AI agents. - Agentforce Builder Considerations
Carefully review the current considerations and limitations for the new Agentforce Builder. - Agents and Data Usage
As an Einstein customer, it’s important for you to know what data is being used by agents.
