Can artificial intelligence move from pilots to production at scale in the enterprise?

This was the central question surrounding Salesforce Inc.’s Dreamforce conference in San Francisco as the customer relationship management powerhouse sought to make it clear this week that its major bet on agentic AI would do exactly that.

At a time when research showing fragmented strategies, weak governance and questionable ROI are limiting AI adoption in the enterprise, Salesforce Chief Executive Marc Benioff (pictured) devoted much of his keynote address today and a follow-up briefing for the media discussing how his firm’s latest releases will bridge what he called the “Agentic Divide.” It all comes back to data and proper access to it, according to Benioff.

“It’s a bifurcation,” Benioff told the Dreamforce gathering. “Your customers are elevated, but what about your enterprise? If you don’t have your data right, you’re not going to get your AI right.”

Leveraging Slack and listening to customers

Salesforce’s announcement this week of Agentforce 360 is clearly designed to streamline the path to enterprise implementation and presumed results across sales, marketing, operations and service. It includes an integrated platform connecting AI agents, Data 360, Customer 360 apps and a conversational interface for humans and agents to work together using Slack.

Much of this week’s releases have been implemented within Salesforce itself, according to Benioff, who noted that 1.8 million conversations on Salesforce Help had already been powered by Agentforce and an automated response tool was calling back 50,000 customers per week.

“We realized at Salesforce that we have to be ground zero,” Benioff said. “Have we been perfect in rolling it out? No. Have we been listening to you? Yes.”

The integration of Slack with Agentforce is a key element in the company’s strategy to bring agents closer to where people work in the enterprise. The update released on Monday included a Slackbot for personalized support and Enterprise Search to retrieve company information using a natural language interface.

It’s an important step in giving autonomous agents more than just an ability to execute tasks. The road to ROI will depend on AI’s ability to meaningfully understand the business and provide correct context, drawing on Slack’s ability to parse conversational meaning.

“Slack has become agentic OS,” Benioff said during his keynote today. “You want to be able to achieve not just prompt engineering, but context engineering. AI alone is not enough. It needs all of these capabilities to connect it, to give it context.”

Slower pace of adoption analyzed

Salesforce provided plenty of customers to tell their stories of AI progress. An executive from Williams-Sonoma described how the firm built and implemented an agent on its website for one of its high-profile brands. In a briefing for the media following the keynote, Athina Kanioura, chief strategy and transformation officer for PepsiCo Inc., described how her company had rolled out Agentforce to 1.5 million stores globally, with plans to be “agent first” by the end of 2026.

Yet even Salesforce’s own customers were surprised that enterprise deployment of AI agents had been slower than expected. “I’m shocked that more people aren’t adopting all of these things because it helps you serve your customer faster,” Williams-Sonoma CEO Laura Alber said during an appearance with the media.

In the media session following his keynote, Benioff addressed questions about the slow pace of deployment. He attributed it to the rapid pace of innovation in the AI field, a process that is often complicated and potentially daunting for firms seeking to work their way through changes in core platforms.

“This is the moment where technology innovation is outstripping customer adoption,” Benioff told the assembled media. “This is architectural changes. This is running their businesses. This isn’t just deploying ChatGPT on my phone.”

Dell Technologies Inc. founder, Chairman and CEO Michael Dell shared a similar perspective in a brief conversation with Benioff during the morning keynote. His company has been adapting to the rapid pace of change as well, with major changes across its product line to reflect the new AI era. Yet, he has also seen other businesses lag behind in AI adoption.

“I saw it as machines thinking for us,” Dell told Benioff. “If we don’t get into this, a new company is going to come along and it’s going to put us out of business. We’ve totally reset the business, reimagined the business. I don’t see as many yet at that stage.”

Moving from product to platform

In addition to questions about Salesforce’s business focus on AI, Benioff was also asked during his meeting with the media to address published remarks over the weekend that he thought National Guard troops should be deployed in San Francisco. The company’s CEO chose not to elaborate.

“I’m going to limit my comments today to Agentforce and Dreamforce,” Benioff said. “We have brought in over 200 of our own police to make sure everyone here has a good experience. We’re doing everything we can to make this a safe experience. This is San Francisco’s largest conference by a factor of 2x.”

Benioff’s focus on Agentforce this week marked the one-year anniversary of the technology’s deployment. Since then, Salesforce has released multiple enhancements, including some major updates within weeks after the previous deployment. It reflected how, like its own customers, Salesforce has moved rapidly to keep pace with developments in the AI field.

“A year ago, Agentforce was a product; now it’s a platform,” Benioff said. “It’s the fastest growing product in our history. I’ve never seen anything grow faster. Because of that we’ve had to go faster.”

Photo: Mark Albertson/SiliconANGLE