AI is set to create new streams of revenue in software, security, identity, and workflows. Yesterday, Salesforce’s strong earnings report boosted the software and SaaS sectors, validating what we discussed back in July: 'CSPs control the horizontal AI foundation, while vertical SaaS controls the business execution layer. Vertical SaaS holds the keys to the “work entry point; business objects, business semantics; historical operational data; system records; industry rules; final action execution; and customer feedback.” It can package model capabilities into high-value business outcomes. This only applies to SaaS platforms that truly own workflows and core proprietary data. Thin applications that simply slap a basic interface onto a general model are more likely to be replaced by model vendors or CSPs.' This time, Salesforce and Anthropic launched ClaudeForce, further proving that the relationship between model vendors and enterprise software isn’t necessarily zero-sum: Claude provides reasoning and natural language entry points; Salesforce provides enterprise data, permissions, business logic, and execution actions; Enterprise software still holds control over system records, business processes, and compliance responsibilities. This indicates that enterprise AI is moving beyond pilot projects and entering the era of 'data integration, agent development, business process invocation, and governance of permissions' in a more engineered and systematic way.