Business leaders at Dreamforce warn AI development outpaces enterprise readiness

At Dreamforce, roughly 50,000 participants highlighted the challenge of matching the rapid pace of AI innovation with practical deployment. Nvidia chief Jensen Huang called for frontier labs to "run as fast as you can," but many executives, including Alec Bronston of Spins and Jaya Rohit Vuyyuru of SummitX, said they are still learning to extract value from models that are a generation or two older. Tim Sanders and Kevin Lee argued that day-to-day AI performance relies more on last year's technology than on cutting-edge releases.

While some firms, such as Databricks, rolled out OpenAI's GPT-6 Astra across their workforce, others like Docusign use a mix of models to choose the best fit for each task. Salesforce announced a revenue target above $63 billion and highlighted its partnership with Anthropic, with up to 1,000 clients testing the Claudeforce beta that connects Salesforce data to Anthropic's Claude chatbot. Overall, businesses remain cautious, often waiting a quarter-year before adopting the newest models.