Salesforce cut 4,000 support jobs for Agentforce — then quietly pivoted back toward deterministic automation

In September 2025, Marc Benioff confirmed that Salesforce had cut its customer support headcount from roughly 9,000 to 5,000, explicitly tying the cuts to Agentforce, its AI agent product: “I’ve reduced it from 9,000 heads to about 5,000 because I need less heads.” It was framed as proof that autonomous AI agents could absorb enterprise support work at scale — Salesforce eating its own dog food.

The walk-back

By December 2025, the story had shifted. Coverage aggregated on a widely discussed HN thread (193 points) — sourced back to reporting from The Information — described Salesforce executives conceding that trust in generative AI had declined internally, with one executive reportedly admitting: “We assumed the technology was further along than it actually was.” A separate story making the rounds, “Salesforce pulls back from LLMs, pivots Agentforce to deterministic automation”, quoted new Salesforce messaging that read almost like a retraction: “While LLMs are amazing, they can’t run your business by themselves… We ground AI in tight guardrails and deterministic frameworks, optimizing LLMs to deliver enterprise-grade reliability.”

HN’s reaction to that framing was blunt. Commenter david_shi noted the new copy “sounds like it’s copy and pasted straight from an LLM,” and commenter PartiallyTyped was more direct: “This reads like a list of obvious things that we have been saying for the past 3 years.” Commenter avereveard went further, speculating “the whole project was just smoke and mirror to justify layoffs.” None of this stopped the layoffs from continuing — Salesforce cut more jobs again in June 2026, with one HN commenter dryly noting “the beatings will continue until morale improves.”

The switching-cost complaint underneath the AI story

Buried in that same 193-point thread is a complaint that predates Agentforce entirely and explains why Salesforce can absorb a rocky AI pivot without losing customers: switching costs. Commenter JohnTHaller put it plainly: “They make hideously complicated software to help businesses manage their business. You need consultants to help integrate it and to make any changes to it. The interfaces are convoluted and require learning how they work rather than having any kind of discoverability… Switching off of their systems is nearly impossible by design.” The AI-agent story and the CRM-bloat story are, in practice, the same story: a platform complex enough to require an AI layer to navigate is also complex enough that customers can’t easily leave.

What it means for CRM buyers

Two things follow. First, if you’re being sold an AI-agent add-on for your CRM, ask specifically what’s deterministic versus probabilistic under the hood — Salesforce’s own reversal suggests even the vendor building it wasn’t sure at launch. Second, if switching cost, not product quality, is what’s keeping you on your current CRM, that’s worth naming explicitly in a procurement review rather than treating as a fixed cost. We track Salesforce Sales Cloud and its self-serve competitors on Crail’s CRM category page, and if HubSpot’s simpler, self-serve model is on your shortlist, our Salesforce Sales Cloud vs HubSpot CRM comparison lays out the trade-offs directly.