Marc Benioff Calls Claims That AI Will Harm Salesforce ‘Crazy Talk,’ Aims for a Billion AI Agents Within a Year

Marc Benioff, the CEO of Salesforce, believes the future of enterprise software is AI agents-or "software bots" that can do some tasks on their own, freeing up those tasks for human workers.
Marc Benioff Calls Claims That AI Will Harm Salesforce ‘Crazy Talk,’ Aims for a Billion AI Agents Within a Year

Marc Benioff, the CEO of Salesforce, believes the future of enterprise software is AI agents-or "software bots" that can do some tasks on their own, freeing up those tasks for human workers.

"What if your workforce had no limits? Now, that's a question we couldn't ask even in the last not only 25 years that I've been doing Salesforce, but the 45 years that I've been doing the software business," Benioff was quoted telling TechCrunch on the Equity podcast.

Already hundreds of customers are using our AI agent platform, Agentforce, in their businesses today," he said. "Our goal is to have a billion agents deployed within one year," he added.

He cited his own experience visiting a Disney park using Disney's Private Tour Guides, a virtual tour guide that scripts the day and handles the skip the line Lightening Lanes.

Oh, my agent just said it looked across all of the flow control of all the rides in the park. It looked at your ride history, of every ride you've ever been on and you haven't been on in Toontown. So let's go that way, he described the experience.

He also talked about healthcare agents doing the type of daily follow-ups to determine whether the patients are following instructions, and he said Gucci is using call center agents, with up-to-the-minute product information, closing sales.

On the other hand, Benioff was quick to pan the agent-making software of his biggest, longtime rival Microsoft. In October, Microsoft introduced 10 new autonomous agents in Dynamics 365 — its Salesforce competitor and an agent-maker as part of Copilot Studio.

In the days following that statement, Benioff publicly trashed Microsoft's products. He compared Copilot to Clippy, Microsoft's much-maligned 1990s talking paperclip cartoon character. He cited research by Gartner about Copilot's flaws, it said, in its opinion, from admin tasks to controlling the agent answers.

Why are they spilling data, and why are users who shouldn't be seeing data are seeing data?" he said of Microsoft's Copilot on Equity. "Copilot has really disappointed so many customers, and not just that, they also have a narrative that they have to write their own LLMs and this and that. And it's not transforming companies.".

Of course, the people who are prognosticating Salesforce's risk from AI — and plenty are doing so these days — aren't arguing that Microsoft's AI agents will outperform.

They are witnessing an extremely seismic change. With GenAI entering the enterprise, enterprise AI startups may create the next multibillion-dollar business that knocks over some of those legacy players. And AI sales rep startups are exploding. So are AI customer service rep startups — two of Salesforce's biggest markets.

More than that, the days of the bad enterprise software interfaces are numbered-only hunting for all the right buttons to do the things. GenAI interfaces should simply mean asking for the sales pipeline report, and the software produces it.

Klarna CEO Sebastian Siemiatkowski made waves in September when he announced his company was abandoning its use of Salesforce and Workday altogether in favor of GenAI apps, some of which are even home grown. This was after they developed a customer service AI assistant called Kiki using OpenAI that is currently being utilised by 90% of its employees.

If OpenAI CEO Sam Altman is to be believed, AGI—basically near-sentient AI that can infer so well it can handle nuanced commands—will be here sooner not later.

Benioff scoffs that any of this is a threat. "I also watched 'Minority Report,' and I also watched 'Her,' and I watched 'Space Odyssey.' I watched them all, but that's not where we are today," he said on Equity.

He is on the ball pointing out that of all enterprise players out there, it is Salesforce that is best positioned to make hay rather than allowing itself to die from AI. After all, AI needs to consume quite a lot of data to become useful, and by virtue of selling solutions that are built entirely for customer relationship management, it has all that data.

"We handle 230 petabytes of data for customers," he said. "You can say that might be one of the biggest things we do for them. And we do it with a security and sharing model," meaning "my banker doesn't know what your account balance is, and your banker doesn't know my bank account balances, although we are in the same bank."

Right now, many companies wonder what OpenAI is doing with data they share with it, even when doing so through ChatGPT Enterprise has security guarantees. Migrating to another cloud provider, like Microsoft, which also sells the latest OpenAI models, requires them to transfer all their data from Salesforce over to Microsoft's cloud. That is an enormous painful project and swaps one big software player for another.

Meanwhile, according to Benioff, Salesforce is using its own agents internally to run its own business as well. "We're using it in our customer service, of course, our call center, we're about to deploy our sales agent. And if you go to help.salesforce.com and you're authenticated as a Salesforce customer, you're now using our agent to get access to all of our help.".

And what about those supposed startups poised to maybe become some sort of thorn in Salesforce's side? Benioff said they may provide Salesforce with M&A targets. After all, Salesforce Ventures is one of the most active corporate venture firms in the land of technology.

"We've done more than 60 acquisitions, and that has been a lot of startups," he said, adding that "even Agentforce" was built with tech from the acquisition of a startup called Airkit.ai, founded by a former Salesforce employee. "We invested in the company. He built the company up. We acquired it, then we re-integrated it into the platform. That's how we were able to release this so fast and get out so quickly."

Blog
|
2024-11-14 18:55:16