Salesforce may have had its ‘ChatGPT moment’ with AI agents – and it will take a lot to knock the company off its perch

Salesforce is excited about AI agents. Very excited. During the opening keynote at Dreamforce 2024, CEO Marc Benioff was effusive about the company’s new Agentforce service, which enables users to create autonomous AI agents.

Agents, attendees heard, are essentially bots capable of automating aspects of a human worker’s role, alleviating workload pressures and streamlining efficiency.

They’re also easily deployable, with users able to build and rollout agents for specific functions within a few clicks.

Enterprises have deployed chatbots and they’ve moved onto AI assistants, or ‘copilots’, but these have thus far failed to truly deliver on the promise of AI automation. For Benioff, agents represent the next step on the evolutionary path of generative AI.

But we’ve heard this level of hype and hyperbole before with regard to AI. For nearly two years now, major vendors have bombarded consumers and enterprises alike with big promises over how transformative this technology will be.

If agents are indeed that next step, that big breakthrough we’ve all been waiting for, then Salesforce’s early start in this new phase could have profound implications for both the future of the technology and the company itself.

Getting that headstart can make all the difference, as history has already shown us.

Rewind to early 2023 and the global technology industry was in a state of pandemonium in the wake of ChatGPT launching. Microsoft and OpenAI struck an early lead in this regard, prompting widespread panic among competitors who then scrambled to keep up.

In January 2023 Google was reportedly in a ‘code red’ situation over what was unfolding at Microsoft, with CEO Sundar Pichai stepping in to reallocate staff and resources to focus heavily on generative AI development.

Other major hyperscaler rivals, including Amazon Web Services (AWS), were in an equally precarious position as the OpenAI train continued to storm its way to dominance.

Microsoft’s success on the back of its lead in the generative AI race has been nothing short of remarkable, helping deliver huge financial gains and surging the firm’s market cap to over $3 trillion at the beginning of this year.

Microsoft itself recently announced a move to launch its own Copilot agents, as did ServiceNow, suggesting that the AI race is going down this route. For Salesforce, as an early starter in this next phase, the outcome could bear similarities to the rip-roaring success Microsoft has had thus far.

Salesforce is going all in on AI agents

This wasn’t a hastily pulled-together announcement or a split-second pivot, and it’s clear that plans for Agentforce have been brewing for some time.

Bola Rotibi, chief of enterprise research at CCS Insight, tells ITPro that Salesforce certainly appears to have “stolen a march” in the domain of agents and shows the company has carefully considered its next steps in the generative AI space.

“I think they’ve caught the zeitgeist of the moment,” she said. “I think what they’ve done is taken a step back and said, okay, what is it that we do and what do people want?”

The hype surrounding generative AI over the last 18 months has been rampant, but there has been a prevailing sense of confusion among many enterprises over how they actually unlock the benefits of the technology, Rotibi noted.

“Generative AI brought something much closer to home and much more accessible, and people could see the value,” she said.

“In that way it kickstarted a race, or the aspiration of what they could think about, but the problem was I think it wasn’t very clear as to where the real value was.”

In curating a service that spans the entirety of its product ecosystem, Salesforce has been able to showcase what AI can do for customers using this approach and provide tangible examples of this service in action.

Throughout Dreamforce, attendees saw demos and examples of AI agents in action, with Benioff and a host of company executives showcasing real-world demos of agents interacting with customers.

“I think they’ve just found a way of basically saying our clients will often use ServiceCloud, SalesCloud, or MarketingCloud, because what they’re also looking for is that integration layer that’s already done for them,” Rotibi said.

“The thing that would’ve become quite obvious [for Salesforce] is that now customers have all this data in these environments, the AI should really be focused on leveraging the data that they have,” she added.

“That’s what people were really crying out for.”

An open-ended, hands-off approach

With Agentforce, there will be no more “DIY AI” for enterprises, as Benioff claimed in his opening keynote. That’s the key differentiator, Rotibi noted, and perhaps the most compelling aspect.

The borderline laissez-faire nature of the service will resonate with customers because Salesforce won’t impose itself on enterprises. The firm appears firmly focused on encouraging customers to unlock value for themselves; if they’re the architect in this scenario, the customer is the interior designer.

“I think what is really good is that he’s [Benioff] not saying that we’re going to create value for you, but there will be something that we’ll do,” Rotibi said.

“We’ll look after the integration, but actually the real value will be down to you and you’ll determine what that is. Agentforce is there to help you augment and provide that insight and visibility, that level of knowledge and intelligence. But it’ll be down to you, you’ll be in control of that.”

The contained nature of Salesforce’s sprawling ecosystem will also calm customer fears over potential data protection issues, which Benioff noted have been a major concern for users of competitor services in recent years.

In a press briefing, the Salesforce CEO gave a withering assessment of Microsoft Copilot and cited a recent Gartner report which suggested the service has been “leaking data all over the floor”.

By keeping enterprise data within its own walled garden, the risk factor is greatly reduced.

“People were concerned about hallucinations because they couldn’t trust it, {the AI model], or it wasn’t based on data that they wanted,” Rotibi said. “The value here is the data that’s inside the organization, in their systems.”

Maintaining this contained approach aligns closely with the values Salesforce have been championing for a while, Rotibi added. The company has placed a strong emphasis on trust, transparency, and an ethical approach to AI that has resonated with customers in recent years.

Looking ahead, Salesforce has the capabilities and the maturity to maintain a significant lead in what Benioff insists is the next phase of enterprise AI.

This week’s announcements have been the culmination of several years of investment and background work to bring it to a point where it can show its card on a global scale.

“Salesforce has bedded itself into what it does well,” Rotibi said. Any moves made by industry counterparts will need to be equally compelling to knock the company off its perch.