In case you missed the chorus of prognostications over the past few weeks, AI-powered agents are set to have a big year. Experts expect generative AI systems that can perform tasks beyond the bounds of a chatbot to come into their own in 2025.
Few companies have bet quite as big on this potential new era as Salesforce. The enterprise giant rebranded its Einstein Copilot platform as Agentforce last year, calling the concept of a copilot “outdated.” CEO Marc Benioff ran an essay in Time Magazine, which he also owns, laying out his vision for an agent-powered “digital workforce” in grandiose terms. “This isn’t just an evolution of technology,” Benioff wrote. “It’s a revolution.”
Salesforce is now rolling out the second generation of its Agentforce platform, which promises a host of new pre-built skills for this “digital labor,” like building marketing campaigns or sales coaching, as well as better reasoning capabilities and data retrieval. Employees will now also be able to talk to these agents in Salesforce-owned Slack.
Agentforce 2.0 will be fully available next month; Accenture, IBM, and Indeed have signed on to the platform, among others.
Sanjna Parulekar, VP of product marketing at Salesforce, told Tech Brew that these agents represent a “step change” from copilot offerings, which she defined as more of an “assistive experience.”
“Can your copilot take action on a few instructions that you give it? Can your copilot proactively reach out to you to do certain things?” Parulekar said. “Is it really functioning autonomously, like a teammate? The answer is no, the answer is absolutely not.”
For example, a manager might build an agent with the capabilities of a model marketing employee, like extensive knowledge of product FAQ and pricing models, and use it to fill gaps a company wouldn’t otherwise have the resources to cover.
Keep up with the innovative tech transforming business
Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.
“On my team, I can never hire enough product marketers to answer seller questions about pricing,” Parulekar said. “I can’t have an army of 50 people that wake up every day doing that; it would be bad business.”
A crowded field: Salesforce is far from the only company thinking about agentic AI right now; Microsoft, SAP, Asana, and others are all offering businesses some form of agent-based product. Parulekar said the guidance Salesforce offers customers and its huge troves of customer data set it apart in this race.
Salesforce is also taking measures to make businesses more comfortable with putting AI agents in charge of their operations. For instance, the company rolled out a testing center for new agents last November.
“Testing these agents is incredibly important to build trust, because you need these ways to test every utterance or every way that a customer might ask a question,” Parulekar said.
No slowing down: The announcement of Agentforce 2.0 comes just months after the first iteration rolled out, and Parulekar said the company has no plans to let up on the pace of updates in the coming year. The ultimate goal is to build out a system of “digital labor” that augments existing teams, she said.
“We now have these agentic systems that can do a lot for us, so let’s put them to work for our highest value tasks that don’t necessarily require a person,” Parulekar said.