Every tech company wants to be seen as the one leading the AI race. Fewer stop to think about what happens when the marketing gets there before the product does.

Salesforce has spent the last year branding Agentforce as a huge step forward, a platform that can essentially run a company's customer experience on its own. Recent reporting says some of that isn't quite true yet. That doesn't mean Salesforce is in trouble, as the deal numbers say otherwise, but there's a real gap between what got promised and what customers are actually getting, and people are starting to notice.

When a company markets something as revolutionary, people expect it to work the moment they try it. Hype can pull people in, but it's proof that keeps them there. Without that proof, even a genuinely strong product can start to look like an overpromise.

Salesforce leaned hard into that word with its Agentforce Commerce release, pitching it as a way to run AI across shopping, service, and customer interactions in one shot. Except Gizmodo and Bloomberg both found that some of the most-hyped features weren't live during testing, including a hospital chatbot most website visitors never even saw. 

There's a University of Chicago Medicine ad built around Salesforce showing a patient chatting with an AI agent that offers to reschedule a delayed appointment. Bloomberg says that feature still doesn't exist, and patients are getting routed to human schedulers like normal.

The AI product itself doesn’t appear fake, some of it just got marketed before it was truly ready to use.

Getting people to sign on the dotted line is one win, but getting them to actually believe in what they bought is a different one. A company can rack up sales while the story behind those sales stays on shaky ground, especially if it's leaning on what's coming rather than what's already there.

You can see that play out in Salesforce's own numbers. The Next Web reported around 29,000 Agentforce deals closed, putting annual recurring revenue near $800 million. Plenty of businesses clearly want in, but Salesforce's stock still dropped about 30% in 2026, which says investors aren't totally sold on AI being enough to shield the company from what's happening across SaaS more broadly. 

So now Salesforce is in an odd spot: selling AI aggressively while the market questions whether it's actually ahead or mostly defending ground it already has.

Slapping AI onto a product doesn't necessarily make it better. The feature only matters if it truly changes something customers care about, whether that's saving time, closing more sales, or making an experience smoother.

Agentforce is central to Salesforce's whole pitch about the future of customer relationships, with humans and AI agents supposedly working side by side across sales, service, and marketing. But between pushback from Gizmodo and the Next Web and plenty of eye-rolling from users on Reddit and LinkedIn, it's fair to ask if Salesforce is selling that future faster than it can build it. One Reddit user summed it up: "Agentforce sales is the goal. So hype-lying about it is the game [...] it's all hype." 

These days, a lot of customers seem less concerned with whether a company uses AI, and more concerned with whether that AI actually does anything for them.

Talking about the future costs nothing, but building it costs everything.

Agentforce's success suggests AI can genuinely drive real business results, but all the criticism around it is a reminder that hype only lasts so long before the product needs to catch up to the pitch. Otherwise, that pitch risks working against you.

I don't think Salesforce is pretending to have an AI product that doesn't exist. They just may have gotten ahead of themselves by sharing the vision before the product was fully ready to match it. Honestly, that's an easy trap for companies chasing the AI wave to fall into: wanting to show people what's coming while still figuring out how to make it work today.

As AI keeps evolving, we'll probably see more companies announce big ideas before they're fully developed. Sometimes that's less about a company being misleading and more about expectations moving faster than the technology. In the long run, the companies that can follow through on their promises, and execute them well, will be the ones that earn people's trust.