Salesforce's Agentforce Faces Vaporware Accusations as Marketing Outpaces Reality

A new Bloomberg report has intensified scrutiny around Salesforce’s Agentforce platform, revealing substantial gaps between the company’s marketing materials and what customers can actually use today. The findings add fuel to ongoing social media criticism labeling the AI agent platform as “vaporware”—a charge that carries particular weight given Salesforce’s position as the prototypical B2B SaaS company.

The Gap Between Demo and Delivery

The Bloomberg investigation centers on several high-profile Agentforce marketing campaigns that showcase capabilities not yet available to customers. The most striking example involves University of Chicago Medicine, featured in a Salesforce advertisement showing patients interacting with AI agents that handle practical tasks: notifying patients when doctors are running late, offering rescheduling options, directing them to available parking spots, booking appointments, and arranging prescription refills.

According to Bloomberg’s reporting, these functions remain unavailable. “Patients who call the hospital system today are greeted with keypad-selection menus and routed to human schedulers,” the report states. The chatbot featured in the advertisement is “still being tested and not visible to most web visitors.” Anonymous sources attributed the delays to “product glitches and difficulty getting sign-offs from internal compliance departments.”

The pattern extends beyond healthcare. Williams-Sonoma presented Agentforce-powered phone features at a stage presentation last year, demonstrating a customer easily purchasing an item they saw online through a synthetic phone operator. Bloomberg notes that “about half a year after that presentation, Williams-Sonoma’s phone line isn’t actually using Agentforce.” The company’s Chief Technology & Digital Officer Sameer Hassan characterized the presentation as showing “possibilities,” with hopes to roll out the features by the holidays.

A promotional video with Finnair depicts similarly aspirational interactions. In one exchange, a customer reports a weather-related flight cancellation, and an agent immediately responds with rebooking options—knowing the user’s itinerary without asking clarifying questions. The seamless experience shown would represent a significant leap from current airline customer service realities.

Salesforce’s Defense and Partial Successes

Salesforce CEO Marc Benioff defended the company’s approach, telling Bloomberg that the scrutinized materials represent “future-oriented marketing.” His response: “Whatever technology we have ever marketed, we have always delivered.”

The Bloomberg report does note evidence of Agentforce delivering value in certain contexts. Appliance and personal tech brand SharkNinja Inc. claims that using Agentforce has resulted in a 20% drop in “service phone calls,” suggesting the platform’s agents can effectively troubleshoot consumer gadget problems. This represents a meaningful, measurable outcome—though notably in a less complex use case than healthcare scheduling or airline rebooking.

Salesforce has also continued expanding its AI strategy beyond Agentforce. The company recently announced “Headless 360,” an initiative inserting “more than 100 new tools” into its CRM system. The goal is enabling users to bypass the software’s graphical interface entirely, letting AI handle operations—positioning this as even easier than the “vibe coding” trend that some argue threatens traditional SaaS models.

The Broader Context: SaaSpocalypse Anxiety

The Agentforce scrutiny arrives amid what some call the “SaaSpocalypse”—the theory that AI-powered code generation threatens traditional SaaS business models. If companies can “vibe code” their own software solutions, the argument goes, why pay for subscription services? Salesforce, as the archetypal B2B SaaS company, faces particular pressure to demonstrate that AI enhances rather than undermines its value proposition.

Salesforce is not alone in facing criticism for AI marketing that outpaces delivery. The report notes that Apple released a 2024 advertisement for a personalized Siri with capabilities that would save actor Bella Ramsey from awkward party situations. Apple subsequently settled a false advertising lawsuit over the ad without admitting fault. The pattern suggests an industry-wide tension between competitive pressure to showcase AI capabilities and the practical challenges of deploying them reliably.

Agentforce was announced in September 2024, meaning the platform has had considerable time to mature. The persistence of gaps between marketing and availability raises questions about whether the challenges are primarily technical, organizational (compliance sign-offs), or reflect more fundamental limitations in current AI agent capabilities for complex, regulated environments.

What This Means for SaaS Teams

For SaaS operators evaluating AI agent platforms—whether Salesforce’s or competitors’—this report underscores the importance of distinguishing between demonstrated capabilities and marketed possibilities. Several practical considerations emerge:

Pilot before committing. The SharkNinja success suggests Agentforce can deliver measurable results in specific contexts. However, the University of Chicago Medicine and Williams-Sonoma examples indicate that complex, regulated, or multi-system integrations may face extended timelines. Request access to test environments and define clear success metrics before enterprise-wide commitments.

Account for compliance overhead. The Bloomberg report specifically mentions “difficulty getting sign-offs from internal compliance departments” as a delay factor. Organizations in healthcare, finance, or other regulated industries should build substantial buffer time into AI agent deployment plans and involve compliance teams early.

Evaluate vendor claims critically. When vendors showcase customer success stories, ask whether the demonstrated features are in production or pilot phases. The Williams-Sonoma example—where a stage presentation showed capabilities not yet deployed six months later—suggests that even named customer references may not reflect current availability.

Consider use case complexity. The contrast between SharkNinja’s success (consumer gadget troubleshooting) and healthcare scheduling challenges suggests that AI agent effectiveness varies significantly by domain complexity. Simpler, more bounded use cases may deliver faster ROI than ambitious multi-system integrations.

One uncertainty worth noting: the Bloomberg report relies partly on anonymous sources regarding the reasons for delays. The relative weight of technical glitches versus compliance challenges versus other factors remains unclear, making it difficult to assess whether these are temporary growing pains or indicators of more persistent limitations.

For SaaS companies building their own AI features, the Salesforce situation offers a cautionary tale about marketing timing. The competitive pressure to announce AI capabilities is intense, but the reputational cost of “vaporware” accusations may ultimately outweigh first-mover advantages—particularly when customers can easily verify whether promised features actually work.