TheChek News Flow

Salesforce’s CEO said this would be the ‘absolute year of Agentforce,’ the name of his company’s AI product. It hasn’t turned out that way, as the enterprise software pioneer faces skeptical customers and greater competition from its peers as well as AI startups.

Salesforce CEO Marc Benioff was upset.

Last fall, one of his technical teams told large Salesforce customers that using Agentforce, the software firm’s new artificial intelligence for automating customer service and other functions, would require extensive planning. The product information the Salesforce team shared with those customers acknowledged the complexity of making the AI perform well.

That message contradicted Benioff, a charismatic salesman who had been telling customers that using the AI was a cinch and they could set up AI agents to handle customer service discussions and other tasks in minutes.

A senior Salesforce leader said that when Benioff noticed the disconnect, he complained to colleagues that the technical team advising large customers, known as Well-Architected, was acting “crazy” and should be fired.

Two months later, after Agentforce became available to all customers, the Well-Architected team was quietly disbanded, with some members taking severance packages and others finding new roles in the company, a former Salesforce manager said. But the team’s insistence that setting up agents was harder than it looked ended up being right, and Benioff was proven wrong: The company has struggled to sell Agentforce in the past year, in part because of the extensive prep work customers need to do to make it work right. And Salesforce in August said it would relaunch the Well-Architected program.

“Many of you told us, clearly and passionately, that it was a mistake” to end the team, two Salesforce executives wrote in a post to customers last month. “You were right.”

Last week, Salesforce issued a less-than-stellar revenue forecast for its current fiscal quarter, raising questions about the quality of Agentforce as well as what kind of money enterprises are willing to spend on it.

The result came despite Benioff’s comments to investors in February that Salesforce was “excited to be…in that kind of rarefied air of delivering a multi-billion dollar AI product line.” Besides customer service, Agentforce also aims to automate sales outreach and responses to IT helpdesk requests.

Benioff isn’t sitting still. When customers found the initial price of Agentforce too high, Salesforce earlier this year lowered the cost per task that the AI handled and also allowed customers to pay in bulk at a discount. After agents created using Agentforce gave inaccurate responses to questions, Salesforce made several acquisitions—including $8 billion for a corporate database firm—and other investments to give the agents better data to work with.

Benioff has found a way out of numerous tough spots since co-founding Salesforce more than a quarter-century ago, but his current circumstances are especially challenging. In turning the entire company’s focus to selling AI agents, Benioff is betting that Salesforce’s deep ties with many of the world’s largest companies, and the fact that these customers are already storing their data in its applications, will make Salesforce the logical choice when those firms decide to adopt AI.

The strategies Salesforce and other software providers have used in the past to drum up interest in new products aren’t working as well for AI products. Those strategies typically include letting customers try out products for free and explaining the business challenges they can help solve.

But this approach hasn’t worked well with AI. While many corporate employees love generalized chatbots like ChatGPT for their own productivity, businesses are taking a more skeptical view of paying for AI after early experiments with the technology. CEOs are increasingly scrutinizing how much money they’re saving by using AI services geared toward enterprises.

“Many of the customers we’re working with are now saying, ‘Let’s pause on the Agentforce conversations,’” said Adam Mansfield, a practice leader at UpperEdge, which helps customers negotiate deals with Salesforce and other software providers. “They’re kind of fatigued by it and they want to talk about getting more out of what they already have.”

Generational Gap

A generational gap could be part of the problem, with tech startups and other young companies incorporating AI into their business operations faster than larger companies.

This is a challenge not just for Salesforce but for rivals such as ServiceNow, SAP, IBM, Oracle, Amazon Web Services and Microsoft, each of which have a large base of enterprise app customers. (The latter three firms have cloud-server businesses that cater to tech startups and other large AI developers, which more than makes up for the challenge involving selling AI apps.)

Further complicating things, these firms are now competing head-on with each other in selling AI agents in ways they didn’t used to. That raises questions about how many of their dueling products will get traction, as customers won’t buy the same type of product from multiple providers.

“The good news for enterprises is that there are so many companies bringing AI products to the table and so they are going to have lots of choices,” said Matt Carbonara, a startup investor who previously worked for Citigroup’s venture arm. “At the same time, being inundated with choices can slow down the process of deciding what to buy.”

Salesforce also finds itself battling startups like Sierra and Decagon that are zeroing in on its largest businesses: customer service and sales management.

Instead of being an “add-on sale” to the products Salesforce already has, Agentforce is “just another competitor” selling agents, said another consultant who works for numerous large Salesforce customers. (And Agentforce has been powered in part by OpenAI technology, likely adding to Salesforce’s costs.)

To Salesforce’s credit, it is one of the few companies to indicate how its agent software is selling, which is a reason Agentforce’s weak sales are garnering attention. And despite chatter sparked by Klarna and other firms last year that said they would use AI to build their own versions of Salesforce apps, that hasn’t happened.

But Benioff may nevertheless have to contend with activist investors, who have watched as Salesforce’s shares have significantly underperformed those of enterprise software peers over the past couple of years.

While many large companies have moved slowly to adopt agents due to data security and privacy concerns, Salesforce bears some responsibility for raising investors’ expectations for Agentforce, said Rishi Jaluria, an equity analyst at RBC Capital Markets.

“It’s really important when companies are coming out with all these new AI features and products that they under-promise and over-deliver,” he said.

Benioff said at one point on the call with analysts that Salesforce “is not a company in crisis,” even though none of the analysts had suggested that.

Benioff has done the opposite. On the earnings call in February, he lauded Agentforce’s “incredible accuracy” and claimed that “customers are really telling us that we’re light years ahead of other providers.”

Benioff Grumbles

Meanwhile, Benioff has to contend with a scrappy new wave of agent startups, one of which is run by someone he knows well.

Bret Taylor, the former Salesforce co-CEO who left in 2022 to co-found Sierra, said recently on TITV that customers are willing to pay higher costs for AI as long as it helps their business.

“You’re seeing this surge of startups right now just because our products are better, candidly, and they work better, and that’s what matters right now,” he said.

For now, Sierra is likely generating less than $100 million in annualized sales, a drop in the bucket for a firm the size of Salesforce. Taylor said he is racing to win deals because it’s a matter of time before customers shift back to buying their software from a provider like Salesforce that offers a suite of enterprise products.

In some internal discussions, Benioff has grumbled about a handful of his employees leaving to join Sierra and taking their customer relationships with them, according to a person who heard the comments. But he and his senior staff are confident that corporate chief information officers will eventually stop flirting with AI startups like Sierra and go back to trusted vendors, said the person.

Benioff has frequently referenced the path of Equinox, which operates a chain of luxury fitness clubs, as a case study for the effectiveness of Agentforce. About eight months ago, Equinox began using Salesforce’s AI agents to help customers find specific types of information on its website and mobile application, such as what time a particular location opens and closes, said Eswar Veluri, the company’s chief technology officer.

Since Equinox doesn’t have a huge volume of customer inquiries, it doesn’t plan to reduce the size of its 20-person “concierge” team, which still handles things like membership cancellations and billing questions, he said. While Equinox uses agents to let customers ask questions and get information about the various classes it offers, customers get routed to human staff when they ask follow-up questions.

Equinox also evaluated agents from Sierra before choosing Agentforce. Veluri said the company was already using Salesforce in other parts of its business, like tracking customer sales leads, and therefore felt it could scale its agent usage faster on Salesforce than it could by using Sierra, which would have required custom software to integrate with its systems, he said. (An Equinox spokesperson declined to comment on how much the company spends on Agentforce and other Salesforce products.)

Agentforce Awakens

As Benioff disbanded the Well-Architected team in late October, Agentforce seemed to get off to a fast start, with 3,000 paying customers by February. That prompted Benioff to declare on an earnings call that month that this would be the “absolute year of Agentforce.”

But on the same call, Amy Weaver, Salesforce’s chief financial officer at the time, cautioned that because Agentforce was still new to many customers, Salesforce expected it to see only “modest” sales in its fiscal year ending in January next year.

It was around this time that some customers began seeing Agentforce give incorrect answers, known as AI hallucinations, as they tested how the software handled customer service inquiries. Another issue was Salesforce’s initial model of charging $2 for every conversation its AI agents handled with customers, which was double the price of some of its rivals.

In May, Robin Washington, the new chief financial officer, said Agentforce had passed $100 million in annual order value, a metric that combines new contracts and renewals. A Salesforce spokesperson said AOV is the same as annual recurring revenue, the value of contracts over the next 12 months.

The spokesperson wouldn’t elaborate on how Salesforce calculated the figure. Agentforce is often bundled with other services such as databases, making it difficult to break out revenue from just that product, according to a consultant who works with Salesforce customers.

While Salesforce has improved the accuracy of Agentforce and the company disclosed 6,000 customers are paying for it, some Salesforce customers are still wary of using it for critical business functions. One Salesforce consultant said that none of the dozen or so consumer retail product companies he works with are using Agentforce for customer service interactions because they aren’t confident that the software can accurately handle tasks such as changing orders to a different shipping location.

Sales Leads Ignored?

Earlier this month, the pressure on Benioff seemed to elicit some memorable comments from him on the conference call with stock analysts.

He said Salesforce was using agents to respond to the 20 million to 100 million people who have reached out to the company over its 26-year history but were ignored because Salesforce didn’t have enough salespeople to respond. (A former senior Salesforce sales executive said Benioff’s comment about the number of customers or potential customers that were ignored wasn’t true.)

Despite Benioff making numerous references to customers using Agentforce, such as DirecTV and Williams Sonoma, he spent more time talking about how Salesforce internally is using the tech to automate more roles. That made some analysts wonder if he was trying to distract from a lack of customer adoption.

Benioff said at one point on the call with analysts that Salesforce “is not a company in crisis,” even though none of the analysts had suggested that.