Meet Agentforce, Salesforce's autonomous AI answer to employee burnout

Salesforce

Autonomous artificial intelligence (AI), or agents that can complete tasks without human supervision, is steadily becoming a part of many enterprises' daily operations. Salesforce is pitching its contribution to the space with a new feature. 

On Thursday, the global CRM platform launched Agentforce, a "suite of autonomous AI agents" designed to take on various tasks across departments like sales, commerce, marketing, and more to scale employee productivity. Formerly called Einstein Copilot, the team of agents, which Salesforce describes as "limitless" in the release, can analyze data and act on customer service concerns, sales leads, and marketing campaign edits. 

Also: IBM will train you in AI fundamentals for free, and give you a skill credential - in 10 hours

Citing its own Trends in AI Report, Salesforce noted a now-familiar area of statistics: employees spend an estimated 41% of their time on "repetitive, low-impact work." The same report found that 65% of desk workers "believe generative AI will allow them to be more strategic." 

Agentforce is intended to be an out-of-the-box feature that can be used across industries and use cases and customized using inexpensive, low-code tools. During a press conference for the release, Salesforce CEO Marc Benioff framed the tool as being accessible for customers who aren't already familiar with the ins and outs of AI technology, calling it "AI for the rest of us."

"In contrast to now-outdated copilots and chatbots that rely on human requests and struggle with complex or multi-step tasks, Agentforce offers a new level of sophistication by operating autonomously, retrieving the right data on demand, building action plans for any task, and executing these plans without requiring human intervention," the company said in the release. 

Benioff positioned Agentforce as proof of AI's potential, in contrast to the billions of dollars businesses have spent on AI thus far without a tangible payoff. He also discussed how Agentforce will address employee burnout -- especially among doctors and nurses -- by providing much-needed support for lower-lift tasks while delivering a reliable customer experience. 

Benioff mentioned the initial release is Agentforce One, with a second generation coming in the future. He added that he hopes to get 1,000 Salesforce customers using Agentforce by next week during the company's Dreamforce conference.

Also: The data suggests gen AI boosts software productivity - for these developers

The release compares Agentforce's tech to that of self-driving cars and explains how Agentforce interprets data to adapt to conditions in real time and can act independently within a company's guardrails. The tool can hand tasks off to humans with summaries of what it's completed so far and recommended next steps. 

Specifically, there are seven types of Agentforce agents:

Service Agent handles customer service and replaces chatbotsCampaign Optimizer uses AI to manage and execute full campaign lifecycles Personal Shopper recommends products and helps with search queries Buyer helps B2B customers find products, make purchases, and track ordersMerchant helps merchandisers do all site-related tasks, from promotions and product descriptions to insightsSales Development Representative (SDR) engages with potential leads 24/7 Sales Coach helps train sales teams with Salesforce data and lets sellers practice pitching Sc

In practice, Agentforce can provide organizations with conversational self-service, answer customer questions using a company database, and direct those with issues to the right resources. According to Salesforce, some customers already using Agentforce, including textbook publisher Wiley, say it outperforms their current chatbots. 

The company says Agentforce is powered by Atlas Reasoning Engine, which is "designed to simulate how humans think and plan." It added that Agentforce demonstrates very low hallucination rates, which it credits to data quality, and comes programmed with toxicity guardrails, which are becoming standard practice in the industry. 

Also: How VR and digital twins are streamlining remote work in technical professions

As part of the launch, Salesforce also introduced Agent Builder, an interface that allows users to customize their Agentforce agents to perform certain tasks. During a demo, Salesforce showed how Saks Fifth Avenue's Agentforce agent, Sophie, made exchanging a sweater more seamless and conversational and how Saks could improve the agent's capabilities even more in the builder. 

Salesforce

"As AI continues to evolve, I'm seeing a fundamental shift in how work is done, with the emergence of agents playing a central role in driving efficiency and scale," said Ritu Jyoti, group vice president and general manager of worldwide AI at IDC, in the release. "Implementing these solutions can boost efficiency and ensure the longevity of organizations in a competitive environment." 

"The only thing we're going to do at Salesforce is Agentforce," Benioff said. 

When asked how Salesforce anticipates Agentforce will affect certain roles (or make them obsolete), VP of Product Marketing Sanjna Parulekar told ZDNET she thinks the technology will create more roles. "We did this exercise on my team to think through, if we had a limitless workforce within marketing at Salesforce, what would we do with it? We thought of 10 different jobs of people that we need on our team." 

Also: Gemini Live is rolling out to all Android users - for free. Here's how to access it

"That really enables us to think through a different way to be a marketing organization. I think each of our customers is going to go through that use case definition: What would you do to create a better customer or employee experience if your workforce size wasn't a limiting factor? Agents are going to help them fill in that white space." 

Agentforce "for Service and Sales," the company clarifies, will be generally available on October 25, with some elements of Atlas Reasoning Engine arriving in February 2025. Pricing begins at $2 per conversation.