Salesforce Scales Back AI Ambitions After Layoffs and Reliability Issues
By TOI Tech Desk | Updated: Dec 25, 2025, 09:47 IST
Salesforce, one of the world’s leading enterprise software companies, is re-evaluating its aggressive AI strategy after encountering significant reliability challenges with large language models (LLMs) and reducing its support staff by approximately 4,000 employees through automation initiatives. Executives at Salesforce have candidly acknowledged that their earlier confidence in generative AI was somewhat misplaced, prompting a strategic pivot toward more predictable, rule-based automation solutions.
From AI Optimism to Pragmatic Automation
Sanjna Parulekar, Senior Vice President of Product Marketing at Salesforce, stated in a recent report by The Information that trust in large language models has declined over the past year. “All of us were more confident about large language models a year ago,” she said, emphasizing the company’s shift away from an AI-first narrative to focusing on deterministic automation frameworks within its flagship product, Agentforce.
This change is significant given that CEO Marc Benioff had previously been a vocal proponent of AI as a transformative tool for Salesforce’s future. However, during a podcast interview, Benioff revealed that the company reduced its support team headcount from 9,000 to about 5,000 — a 44% cut — owing to the deployment of AI agents, underscoring the extent to which automation is reshaping Salesforce’s operations.
Technical Issues Shed Light on AI Limitations
Problems with LLMs in real-world applications have revealed critical shortcomings. Muralidhar Krishnaprasad, CTO of Agentforce, explained that when models are given more than eight instructions, they begin to drop some, a severe issue for business processes requiring precision.
A notable example came from Vivint, a home security firm serving 2.5 million customers. Vivint uses Agentforce to manage customer support, but they experienced failures in sending out customer satisfaction surveys after interactions, despite clear instructions. Salesforce collaborated with Vivint to implement "deterministic triggers," ensuring consistent execution and addressing the AI’s unreliability.
Additionally, Salesforce executive Phil Mui highlighted the phenomenon of AI "drift," where AI agents lose focus if users pose irrelevant or off-topic questions. For instance, a chatbot designed to assist with form completion might get sidetracked by unrelated queries, compromising task accuracy.
CEO Benioff Reassesses AI’s Role
The retreat from large language models represents an ironic turn for Marc Benioff, who had earlier described AI transformation as central to Salesforce’s future. In recent comments to Business Insider, Benioff acknowledged the limitations of AI models, particularly their tendency to generate "hallucinations" without solid data grounding. Instead, he emphasized the importance of robust data foundations over solely relying on generative AI.
Benioff even floated the possibility of rebranding Salesforce as "Agentforce," reflecting a growing focus on their deterministic AI platform. However, this marketing enthusiasm contrasts with the technical hurdles the company is now facing.
Financial and Market Implications
Salesforce’s stock price has fallen roughly 34% from its December 2024 peak, signaling investor concerns about the company’s AI strategy and broader market challenges. Nevertheless, Agentforce remains a promising revenue stream, with projected annual earnings exceeding $500 million.
Given Salesforce’s extensive enterprise customer base relying on AI-powered solutions, this partial withdrawal from LLMs may have wide-reaching effects as companies reassess the gap between cutting-edge AI innovation and practical business reliability.
Salesforce’s Commitment to Trusted AI
A spokesperson for Salesforce stressed the company’s ongoing dedication to delivering AI that enterprises can depend on. “While LLMs are amazing, they can’t run your business by themselves. Companies need to connect AI to accurate data, business logic, and governance to turn the raw intelligence that LLMs provide into trusted, predictable outcomes,” the spokesperson said.
“This is why we built Agentforce: trusted AI infrastructure that drives real business value. We ground AI in tight guardrails and deterministic frameworks, optimizing LLMs to deliver enterprise-grade reliability. Trusted. Reliable. Secure. This is what AI is meant to be.”
About TOI Tech Desk
The TOI Tech Desk is a dedicated team delivering timely and authoritative technology news for The Times of India. Covering topics from AI and cybersecurity to gadgets and digital trends, the team ensures readers receive accurate and insightful updates on developments shaping the tech world.
For more tech news and updates, visit TimesofIndia.com/tech-news
