Salesforce Agentforce: What you need to know

Since its launch at the Dreamforce conference in September 2024, Salesforce Agentforce changed the conversation around AI, customer experience and customer service. Shortly after the launch, Salesforce announced it was hiring more than 1,000 employees to meet the demand for Agentforce. 

At the time of the Agentforce announcement, Salesforce CEO Marc Benioff called Agentforce “the third wave of AI — advancing beyond copilots to a new era of highly accurate, low-hallucination intelligent agents that actively drive customer success.”

The platform is so important to Salesforce that Benioff declared at the Agentforce launch, “The only thing we’re going to do at Salesforce is Agentforce.”

What is Agentforce?

Agentforce is essentially a platform layer in the Salesforce ecosystem. Agentforce bots are generative AI bots, but unlike popular genAI platforms like ChatGPT and Google Gemini, Salesforce says its Agentforce bots can be trusted to take action on their own.   

Unlike earlier chatbots, AI agents do not follow a prescribed pattern of answers to common queries, but are capable of responding more flexibly to questions or issues because they are trained on internal data and are capable of drawing on the support of a large language model (LLM). Salesforce’s agentic AI, for example, can be trained on data in Salesforce Data Cloud.

Agentforce combines three major Salesforce tools — Agent Builder, Model Builder and Prompt Builder — to deliver out-of-the-box bots that work across industries. The bots are customized using inexpensive, low-code tools.

The initial agents created with Agentforce include:

• Campaign Optimizer, which manages and executes full campaign lifecycles. 
• Service Agent, which replaces chatbots in handling customer service.
• Buyer, which helps B2B customers find products, make purchases and track orders.
• Personal Shopper, which recommends products and helps with search.
• Merchant, which does all the site-related tasks for merchandisers, including promotions, product descriptions and insights.
• Sales Development Representative, which engages with potential leads 24/7.
• Sales Coach, which helps train sales teams and lets sellers practice pitching.

General availability for Agentforce One began on Oct. 25, 2024, with pricing starting at $2 per conversation, with discounts available to large users

Salesforce also announced the Agentforce Partner Network at the launch. It lets other organizations deploy third-party agents or use third-party actions when designing custom agents in Agentforce.

Dig deeper: Customer experience management in the age of agentic AI

How does Agentforce work?

During the initial Agentforce announcement at Dreamforce, Benioff said Agentforce is designed for customers who don’t know much about AI or building bots.

“They are not going to have to be an expert in an LLM, they are not going to have to understand all of these deep capabilities that you would need to know to be a computer scientist,” he said. This is “AI for the rest of us.”

Agentforce is powered by an upgraded version of Salesforce’s Einstein AI called Atlas Reasoning Engine, which is “designed to simulate how humans think and plan.” 

At the announcement, Salesforce drew a comparison to self-driving cars. Agentforce can interpret data to adapt to conditions in real time and can act independently within a company’s guardrails. 

Benioff said Agentforce has the lowest hallucination rate of any generative AI but did not say what that rate is. 

Enhanced Atlas Reasoning Engine

Salesforce announced in December 2024 that its Atlas Reasoning Engine was improved with advanced reasoning and retrieval augmented generation (RAG), allowing agents to handle more complex, multistep questions and provide more accurate and precise answers.

The enhanced RAG capabilities let users verify information by giving inline citations with agent responses. Citing the exact sources builds trust in the AI’s output. It also helps users check for accuracy. 

Agentforce integrations with other Salesforce tools

Data Cloud

Data Cloud provides the customer data that grounds Agentforce, making agents more contextually aware, knowledgeable and adaptable to customer needs. Data Cloud connects, unifies and harmonizes customer data and metadata, giving agents access to the exact information they need to deliver precise, contextual responses. It makes structured and unstructured data – including emails, transcripts and PDFs – easily digestible by large language models (LLMs). The RAG functionality allows users to upload files and URLs as data sources for their agent. 

Customer 360

Agentforce is deeply integrated with Salesforce Customer 360, leveraging applications like sales, service, marketing and commerce. A complete view of the customer is critical to seamless hand-offs to human agents with the full conversation history.

MuleSoft

MuleSoft is an integration, automation and API management platform. It extends Agentforce capabilities by enabling Salesforce developers and admins to leverage their APIs and bring in third-party data.

Slack

Slack is a central point for employees to interact with Agentforce. They can access information, trigger workflows and manage tasks directly within Slack, which means they don’t have to jump between applications.

Agentforce also uses Slack Enterprise Search to learn from conversations in Slack channels and direct messages. This helps agents provide more relevant responses and actions based on the context of conversations.

Tableau 

Tableau Einstein is an AI-powered visual analytics platform that includes Agentforce. It is designed to intelligently embed data and insights across the flow of work and encourages employees to use data to drive action. 

Dig deeper: 4 key features in Salesforce’s Agentforce 2.0

Agentforce Testing Center

In December 2024, Salesforce announced The Testing Center, which lets businesses test their AI agents before deployment. The Testing Center gives Agentforce users an opportunity to  ensure the agents follow instructions, stay factual and work quickly. This helps mitigate risks and ensures agents perform consistently and reliably. It also provides tools that make it easier to test, audit and scale agents.

Library of skills and integrations

Also announced in December 2024, this library spans key business applications like CRM, Slack and Tableau, and integrations from AppExchange partners. These ready-to-use skills simplify the process of building tailored AI agents. The library has pre-built Slack Actions like Create Canvas and Message Channel, which let teams create agents for their Slack workflows.

Here are some examples:

• The CRM Skills let you create agents to automate and enhance customer experiences. These agents can do things like lead nurturing, sales coaching, and scheduling service engagements among other functions. 
• The Sales Development skill creates agents that nurture leads based on predefined rules. 
• The Sales Coaching skill lets agents actively participate in sales calls, providing real-time feedback to representatives and improving their performance.
• The MuleSoft integrations let companies use Agentforce across their entire business ecosystem. For example, it can connect Agentforce to its ERP system to automate order fulfillment processes. It can also integrate with marketing automation platforms to trigger personalized email campaigns based on customer interactions.
• The AppExchange boosts Agentforce capabilities by letting users access specialized skills created by partners.

Which companies are using Agentforce?

Salesforce itself is using Agentforce internally. It says agents at help.salesforce.com now handle 83% of customer support queries independently, with human escalations dropping by 50% in the two weeks after it was implemented.

RBC uses Atlas and Data Cloud with Agentforce to help its financial advisors. Salesforce Data Cloud integrates all RBC client data into a unified system, giving advisors a comprehensive view of their clients. Atlas can answer complex client questions using multistep reasoning. It also pulls from RBC’s data and business logic to deliver a reliable answer.

Accenture uses Agentforce to give employees quick updates on important accounts. It helps automate document retrieval and makes recommendations for solving business problems. It also streamlines proposal creation and visualization and enhances employee collaboration by providing real-time updates, retrieving documents, offering proactive recommendations and creating proposals. 

Other companies publicly identified by Salesforce as Agentfore customers include:

• The Adecco Group
• BACA Systems
• OpenTable
• Saks
• Wiley

Dig deeper: SharkNinja embarks on its Salesforce AI journey