Is the world ready for autonomous AI? Salesforce CEO Marc Benioff makes the case for agents

Marc Benioff wasn’t buying the premise of my question.

Recording an upcoming GeekWire Podcast episode with the Salesforce CEO (coming this weekend), I started by asking, essentially, whether the industry is moving too quickly beyond the concept of artificial intelligence as a work companion, leaping headfirst into a world where AI agents act autonomously.

“Well, I think you’re taking a pretty big jump into the abyss there,” he began.

Benioff urged me to take a step back, pointing out that Salesforce has been working on AI for more than a decade through its Einstein platform. The rise of generative AI, connected securely to corporate data, opens the door to new forms of AI that can reason, plan, and take action on behalf of a business.

“It gives us the ability to achieve capabilities with enterprise software that we just have not been able to do before,” he said. “We are in the zone.”

Years from now, people will look back on this week as the moment that AI agents started to enter the mainstream — or jumped into an abyss of their own.

In addition to this week’s general availability of Salesforce Agentforce for sales and service, Amazon-backed Anthropic unveiled a new experimental API that lets Claude control computers autonomously. Microsoft also announced its own AI agents as part of its Copilot platform, looking to preempt the Salesforce launch.

“This new world is being defined by a rich tapestry of AI agents, which can take action on our behalf, including personal agents across work and life, business process agents, and cross-organizational ones,” wrote Microsoft CEO Satya Nadella in his annual letter to shareholders Thursday afternoon.

He added, “These agents will be able to work in concert as a new input to help make small businesses more productive, make multinationals more competitive, make the public sector more efficient, and improve health and education outcomes broadly.”

Benioff has been outspoken in his criticism of Microsoft Copilot in recent weeks, saying that it has disappointed users, put their data at risk, and generally given enterprise AI a bad name, which we’ll discuss on the upcoming podcast.

But on the potential for AI agents in general, he and Nadella seem to be aligned.

“I would say I’ve never been more excited about anything in my entire career,” Benioff said. “We can really see that software can do things that it’s never been able to do before … It can improve our margins, grow our revenues. It can improve our KPIs, fundamentally improving our overall business performance.”

He cited a few different examples to illustrate the potential of Agentforce:

• Book publisher Wiley was able to scale its sales, service, marketing and customer outreach, augmenting human workers to expand its capabilities during the back-to-school textbook season.
• Saks Fifth Avenue was able to build a system to manage returns and customer service using Agentforce in one hour during Salesforce’s Dreamforce conference, and it’s in use on the Saks site today.
• In the realm of healthcare, Benioff said AI agents can help patients by reminding them to take follow-up actions after medical procedures, like drinking water, taking medications, and scheduling appointments.

Agentforce uses generative AI to move beyond the rule-based approach of traditional chatbots, which relied on complex dialog trees and if-then logic. Compared with rivals, Salesforce says it offers a more integrated experience, grounded in customer data, with built-in guardrails and safety measures.

“This is our dream of why we all got into the software business, because we thought this is where software was going,” Benioff said. “We are at this moment, right now.”

Listen to the GeekWire Podcast with Salesforce CEO Marc Benioff starting Saturday, Oct. 26. Subscribe in Apple Podcasts, Spotify, or wherever you listen.