Salesforce AI gains 6,000 customers while quietly pulling back on large language models

Salesforce AI Platform Shows Explosive Growth Amid Industry Skepticism

While debates about an AI bubble intensify across Silicon Valley, Salesforce AI has quietly demonstrated that enterprise AI adoption is accelerating at a remarkable pace. Agentforce, the company's autonomous AI agent platform, added 6,000 new enterprise customers in a single quarter—a 48% increase that brought its total customer base to 18,500, up from 12,500 the previous quarter 1

1

. These customers now run more than three billion automated workflows monthly, and the platform has processed over three trillion tokens, positioning Salesforce as one of the largest consumers of AI compute in the enterprise software market 1

1

.

The company's agentic product revenue has surpassed $540 million in annual recurring revenue, a figure that Madhav Thattai, Salesforce's Chief Operating Officer for AI, described as "pretty remarkable for enterprise software" given that these products have only been available for a couple of years 1

1

. This growth arrives as venture capitalists and analysts question whether billions pouring into AI infrastructure will generate proportionate returns, with Meta, Microsoft, and Amazon committing tens of billions to AI development.

Trust Issues Force Strategic Pivot Away from Large Language Models

Despite the impressive customer growth, Salesforce executives now openly acknowledge serious trust issues with generative AI technology. Sanjna Parulekar, Senior Vice President of Product Marketing, admitted that "all of us were more confident about large language models a year ago" 2

2

. The company is now reducing its reliance on large language models after facing repeated reliability issues during customer deployments.

Muralidhar Krishnaprasad, Chief Technology Officer of Agentforce, revealed that the models begin failing when given more than eight instructions, causing them to drop some directives—a serious risk for businesses requiring precision and consistency 2

2

. Home security firm Vivint, which uses Agentforce for customer support serving 2.5 million customers, experienced problems where the system failed to send satisfaction surveys despite clear instructions 2

2

.

Salesforce is now prioritizing predictable, rule-based automation over open-ended generative AI to remove "the inherent randomness of large models" 2

2

. This represents a significant retreat from the AI-first messaging that dominated the tech industry just months ago.

Workforce Impact and the Human Cost of AI Adoption

The deployment of AI agents has already reshaped Salesforce's workforce structure. CEO Marc Benioff revealed that the company reduced its support staff from 9,000 to approximately 5,000 after deploying AI agents—effectively laying off around 4,000 employees 2

2

. Benioff stated bluntly: "I've reduced it from 9,000 heads to about 5,000, because I need less heads" 2

2

. These layoffs underscore how quickly AI tools have moved from experimental features to systems directly influencing staffing decisions, even as questions about their reliability persist.

Why Enterprise AI Trust Has Become Critical for CIOs

The distinction between AI experimentation and AI adoption at scale comes down to trust, a word that appeared repeatedly in interviews with Salesforce executives and industry analysts 1

1

. Dion Hinchcliffe, who leads the CIO practice at The Futurum Group, said the urgency around enterprise AI has reached unprecedented levels. "I've never seen anything like this before. In my entire career, I've never seen this level of business focus—boards of directors are directly involved, saying this is existential for the company" 1

1

.

CIOs now face intense pressure from board members demanding answers about how their companies will avoid disruption by AI-native competitors 1

1

. Yet this creates a paradox: companies want to move fast on AI, but the autonomy that makes AI agents valuable also makes them dangerous. An agent that can independently execute workflows and process customer data can also make mistakes at machine speed or be manipulated by bad actors.

Technical Challenges: Drift, Hallucinations, and Governance Requirements

Salesforce executive Phil Mui identified AI "drift" as a significant problem in an October blog post 2

2

. When users ask irrelevant questions, AI agents lose focus on their main task—for example, a chatbot designed to guide form completion may get distracted by unrelated customer questions, reducing effectiveness in structured business workflows.

Marc Benioff now cites concerns around hallucinations when AI systems operate without proper data context 2

2

. He revealed that he is drafting Salesforce's annual strategy document with data foundations, not AI models, as the top priority. Building production-grade agentic AI systems requires hundreds of specialized engineers working on governance, security, testing, and orchestration—infrastructure most companies cannot afford to build themselves 1

1

. According to Hinchcliffe, "the average enterprise-grade agentic team is 200-plus people," while Salesforce has over 450 people working on agent AI 1

1

.

Market Implications and What Comes Next

Salesforce shares are down approximately 34% from their December 2024 peak, despite the company expecting Agentforce to generate more than $500 million in annual revenue 2

2

. Benioff even suggested the company could one day rebrand itself as "Agentforce," telling Business Insider "that would not shock me" after learning from focus groups that customers are tired of hearing about cloud computing 2

2

.

The Futurum Group's recent analysis of agentic AI platforms ranked Salesforce slightly ahead of Microsoft as the market leader 1

1

. Yet Salesforce's decision to scale back large language models reflects a broader industry moment, as companies attempt to bridge the gap between AI hype and dependable, day-to-day business use. Watch for how other enterprise AI platforms address similar reliability issues, whether the shift toward rule-based automation becomes an industry standard, and how workforce reductions impact long-term customer support quality.