Introduced in October 2024, Agentforce offers a suite of preconfigured autonomous AI agents and the tools for businesses to build their own.
Salesforce - the vendor behind Agentforce - believes the platform represents the "third wave of AI" - going beyond "copilots" to introduce a new era of accurate, multifunctional AI agents.
These agents integrate AI across customer-facing functions and beyond into the broader enterprise.
In doing so, they automate specific tasks but can collaborate to mechanize multi-step flows.
For instance, one may anticipate customer needs and work with another that takes proactive actions based on that intelligence.
That's the big differentiator of agentic AI platforms like Agentforce. They don't just surface insight; they trigger actions on the back of that insight.
Additionally, Agentforce empowers anyone to create custom agents with a no-/low-code user interface.
But that's just the beginning. Here’s everything you need to know about Agentforce.
What Is Agentforce?
Agentforce is Salesforce’s agentic AI platform. It allows users to build, customize, and deploy autonomous AI Agents capable of supporting employees and customers.
Organizations can use Agentforce to deploy pre-built agents or create their own with out-of-the-box templates and familiar Salesforce tools, such as Flows, Prompt Templates, Apex, and the Data Cloud.
Agentforce agents come with the Einstein Trust Layer for data protection, offer access to customizable guardrails for enhanced controls, and - if integrated with Data Cloud - are grounded in the business's data.
Additionally, agents don’t require pre-determined conversational flows (like Einstein Copilots). Instead, they can adapt in real time to changing requirements, using intuitive, cutting-edge language models and Retrieval Augmented Generation (RAG).
At the heart of Agentforce is the all-new "Atlas Reasoning Engine". Built on a proprietary system designed to simulate how human beings think, Atlas reasons through problems.
Additionally, this engine dynamically evaluates user queries, refines them, and retrieves only the most relevant data for more reliable performance.
Lastly, since Agentforce integrates seamlessly with all of Salesforce’s apps, including its Customer 360 apps, MuleSoft, Apex, and Salesforce Flow, customers can extend the functionality of each agent with pre-optimized workflows and actions.
How Agentforce Works: The Basics
Companies can leverage preconfigured agents within Agentforce or create their own within the Agent Builder inside Agentforce Studio.
To do so, a business may use natural language – which simply means "describing" what it wants the agent to do.
So, the user may first describe the agent's purpose and define the knowledge and data sources it can access.
Businesses may group these sources by "topic" to make this simpler. For example, if a creator builds an agent for customer service, it can tag the agent with all topics relevant to customer service.
These topics may have specific actions tied to them for inspiration. Yet, brands can also create new actions based on Apex data, Flows, Mulesoft APIs, and prompts to fully customize their agents.
Once they've created an agent, a user can test-drive it, learn, optimize, and then embed the agent within their tech ecosystem.
When deploying an agent, it will automatically function across the Salesforce apps listed in its description, such as in the Service or Sales Cloud.
If connected to Data Cloud, Agentforce agents will dynamically update themselves too, as new data enters the ecosystem.
The Preconfigured Agentforce Agents
The great thing about Agentforce is that a user doesn’t need to be an AI expert to create a custom agent.
Indeed - alongside all the necessary building blocks to design sales, service, and other types of agents – are agent prototypes that businesses can customize without writing code.
Salesforce released 100 of these preconfigured agents when launching the solution. Some of the standout out-of-the-box agent options include:
