At Dreamforce 2026, Salesforce introduced AIforce, a strategic umbrella brand that signals a fundamental shift in how enterprise software is consumed. Rather than forcing users to navigate traditional dashboards, Salesforce is effectively turning the agent into the user interface. By bringing data, workflows, and business logic directly into external environments like Claude, Slack, and Lightning, the company is betting that the future of work happens where the user already is, not where the database lives.
Marc Benioff framed this as an Salesforce’s AIforce announcement: “AI is creating an interface revolution. We are combining model intelligence with all the context that customers have built into Salesforce to create an intelligent, dynamic, composable system that is securely governed, built with Zero Data Retention, and designed to work with the core systems that already run your business.” The architecture behind this is the Headless Toolkit, which exposes Salesforce elements via APIs and Model Context Protocols (MCPs). This allows for a seamless experience where users don’t need to migrate data or learn new permission models; existing governance and security settings carry over automatically. The company reports that 100,000 users activated the Agentforce Coworker within its first 35 days, suggesting a rapid appetite for this embedded approach.
AIforce is structured into three primary components. Claudeforce integrates Salesforce directly into Anthropic’s Claude, offering 37 prebuilt sales skills such as meeting preparation, deal health reviews, and automated pipeline updates. “Salesforce in Claude brings this same frontier intelligence into the systems where much of the world’s commercial activity happens,” said Dario Amodei, CEO and co-founder of Anthropic. Slackforce positions Slack as the central hub for the Agentic Enterprise, featuring interactive dashboards, a personal AI teammate, and native CRM capabilities. Finally, Agentforce Coworker provides the same intelligence directly within the Salesforce Lightning interface. Crucially, the entire system operates under a Zero Data Retention policy, meaning customer data is neither stored nor used to train external models.
This shift toward an agent-as-UI model creates a significant tension in enterprise pricing. For decades, the SaaS industry has relied on the per-seat model, which is predictable for vendors but often disconnected from actual usage. Salesforce is now attempting to bridge this gap by offering four parallel buying routes. These range from a $5 per-user monthly fee to comprehensive editions starting at $550 per user. However, the most disruptive option is the pure consumption model, which charges $2 per conversation or $500 per 100,000 Flex Credits, bypassing per-user fees entirely.
The coexistence of these models is a calculated move to capture different segments of the market. By offering consumption-based pricing, Salesforce is acknowledging that as agents take over more tasks, the traditional seat becomes a less accurate measure of value. If an agent is doing the work of a human, should the company pay for the human’s license, the agent’s activity, or both? For a 30-seat enterprise team, the realistic first-year total cost of ownership — factoring in the required Service Cloud Enterprise Edition, implementation, and credits — can range from $200,000 to over $450,000.
The practical caveat here is that while the agent-as-UI sounds frictionless, the cost structure remains complex. Enterprises must now decide whether to commit to fixed per-user costs or embrace the volatility of consumption-based billing. While the former provides budget certainty, the latter aligns costs more closely with the actual volume of automated work. As companies scale their use of agents, they will likely find that the front door to their enterprise is no longer a single application, but a distributed network of AI interfaces that carry a variable price tag.
Ultimately, the success of AIforce will depend on whether these agents can reliably handle the complexity of enterprise workflows without requiring constant human intervention. The technology is clearly designed to remove the friction of switching between apps, but it introduces a new layer of financial management. Decision-makers are no longer just buying software seats; they are buying a metered service that scales with the intelligence of their operations. Whether this leads to greater efficiency or simply a new, more opaque category of IT spend remains the central question for the coming year.