- Salesforce has expanded its AI tools with Agentforce, a platform to help users automate tasks.
- Agentforce uses generative-AI models that build on Salesforce’s existing products.
- This article is part of “CXO AI Playbook” — straight talk from business leaders on how they’re testing and using AI.
For “CXO AI Playbook,” Business Insider takes a look at mini case studies about AI adoption across industries, company sizes, and technology DNA. We’ve asked each of the featured companies to tell us about the problems they’re trying to solve with AI, who’s making these decisions internally, and their vision for using AI in the future.
Salesforce is best known for its customer-relations-management software, which it says is used by over 150,000 companies, including Amazon and Walmart. It also owns Slack, a popular communication platform.
The company is expanding the availability and capabilities of its AI agents with Agentforce, a platform that helps Salesforce clients build and deploy agents to automate tasks such as generating written reports from sales data and creating summaries of conversations in Slack.
Situation analysis: What problem were they trying to solve?
AI agents aren’t new to Salesforce. In 2016, the company released Einstein, an AI service for its CRM software. Einstein handled some of the tasks Agentforce can tackle, but the actions it could take were more scripted.
With the arrival and rapid progress of generative-AI models, Salesforce saw an opportunity to improve agents’ decision-making and understanding of natural language inputs, like email.
That initially led to Einstein GPT, a generative-AI overhaul of Einstein. Then it evolved into Einstein Copilot, announced in early 2024. Now it’s called Agentforce, which offers prebuilt and customizable agents.
“We had to recognize that our customers want to either extend the agents that we ship or they want to build their own customer agents,” Tyler Carlson, Salesforce’s vice president of business development, told Business Insider.
Key staff and partners
Agentforce’s core technology, which Salesforce calls the Atlas Reasoning Engine, was developed by the company’s engineering team. Salesforce also allows its clients to access AI models from third-party technology partners including OpenAI, Anthropic, Amazon, and Google.
Salesforce has worked with internal and external partners to showcase its AI agents. The company uses Slack to spotlight AI agents that can be accessed directly from the communication platform. (Currently, Agentforce’s integration in Slack is in beta.)
“The reason Slack is interesting for these employee-facing agents is that it surfaces these automations and capabilities to users where they’re spending their time,” Carlson said.
AI in action
Salesforce says the Atlas Reasoning Engine uses what’s known as ReAct prompting to help agents reason through problems.
ReAct helps an AI agent powered by a large language model break down a problem into its components and then provide steps to solve it. After each action, the AI agent evaluates the result to judge whether it has satisfied the user’s prompt. If not, the agent loops through different reasoning and action steps until it reaches a satisfactory result.
Salesforce says this can lead to fully autonomous agents that offer personalized support based on a user’s specific inquiry and take action — like scheduling an appointment or meeting — without human intervention.
The concept is similar to OpenAI’s new o1 model, which uses the chain-of-thought prompting technique to work through prompts step-by-step.
The Atlas Reasoning Engine is proprietary technology built by Salesforce, and it has access to the company’s internally developed LLMs. But it’s also designed to work with LLMs provided by partners — the company expects many agents will take advantage of that capability.
“This is what I find most interesting about agentic systems,” Carlson said. “They’re not monolithic. They’re not a single LLM.”
Allowing access to third-party LLMs can be risky because it means passing data to an outside vendor. To address this, Salesforce instituted a policy that prohibits data retention. It also provides safeguards designed to identify and reject inappropriate responses to prompts.
Salesforce’s clients can also use a tool called Agentbuilder to create and deploy custom Agentforce agents that can act on triggers, such as receiving an email or making a sale, without a direct prompt from a company’s employees. An automated agent might open messages from a general email inbox and forward them to specific departments based on the agent’s understanding of the message’s intent.
Agentforce also uses retrieval-augmented generation to create agents that can answer questions based on a company’s internal documents and private data. For example, this year Salesforce and Workday created an “AI employee service agent” designed to answer employee questions about a company’s HR policies.
Did it work, and how did they know?
Carlson said Salesforce measured the success of its software by the success of its clients, adding that the results so far had been positive. “That includes promising statistics, like seeing 90% of customer inquiries resolved by an agent,” he said.
Salesforce will also judge its success by Agentforce’s adoption as a platform. The company wants to see clients broadly adopt AI agents designed with Agentbuilder and deploy them within their CRM systems and in Slack. It also wants to see the new AI agents take on larger caseloads and resolve interactions to satisfy clients and customers.
“By next year I want to see a larger and more open ecosystem of partners,” Carlson said. “I want Agentforce to have thousands upon thousands of agent skills, topics, and a lot more that our customers can take advantage of.”
Read the full article here
