"Agents don't need dashboards". The line, delivered by Vala Afshar, Chief Digital Evangelist at Salesforce, has meaningful implications for enterprise software systems. In an interview with CNBC-TV18, Afshar offered insights into Salesforce’s headless enterprise strategy, highlighting the shift from human-operated software interfaces toward systems increasingly designed for AI agents to execute workflows directly.
At a Slackbot event on 31 March, Salesforce co-founder, Parker Harris, ask a room of press, analysts and partners: "Why should you ever log into Salesforce again?”. Harris continued: “Maybe you never will. Maybe you will go into Slack". What he meant by this is that Slack can be used as a front door to Salesforce’s CRM. Salesforce’s CEO, Marc Benioff, confirmed the company was rebuilding its entire user interface around Slack. The formal announcement of Headless Agentforce 360 at TDX followed shortly after.
The Rise of the Headless Enterprise
Informatica from Salesforce has pursued a parallel direction in enterprise data management, becoming what it describes as the first platform to expose every data management capability as a reusable, governed service that any AI agent can invoke instantly, with native MCP support. The platform integrates directly with Claude, Slackbot and Cursor, reinforcing the shift toward agent-driven data access over human-operated tooling.
Together, the developments point toward the rise of the headless enterprise, where APIs, orchestration layers and governed data infrastructure matter more than traditional software interfaces. For human users, that may mean navigating enterprise systems through conversational tools like Slack rather than dedicated applications. For AI agents, no interface is needed at all.
According to Afshar, Salesforce now operates 60 MCP servers providing programmatic access to its full platform, meaning agents can execute workflows, update records and resolve cases without ever opening a graphical interface.
Salesforce's Broader Agentic AI Vision
Afshar was explicit about what he sees as the competitive stakes, arguing that businesses not yet operating as agentic enterprises are effectively on a path to irrelevance. He framed this as "digital Darwinism": the idea that survival now depends not on scale or heritage, but on speed of AI adoption. The pressure is already visible in how agentic AI is reshaping customer experience strategies across industries.
Afshar shared some of figures around internal operations, helping to illustrate the scale involved. Agentforce has resolved more than 3.5 million support cases in the past 16 months. Across its 85,000-strong workforce, 60,000 employees use agents daily through Slack. Internally, Salesforce manages 300 active agents, each monitored against performance metrics and ranked by productivity. Agents that underperform are removed, while those that exceed expectations receive additional data, rules and capabilities. This performance management language is strikingly similar to that applied to human employees.
The Salesforce executive reframed the concept of hybrid work, describing it no longer as a question of where employees choose to work, but as humans and AI agents working together to create value at the point of need. This aligns with Salesforce’s shift towards positioning agents as operational partners, rather than simply background tools.
From SaaS Applications to Agent Operating Systems
Afshar commented on Salesforce’s move away from pure per-seat licensing toward a hybrid model that includes consumption-based and outcome-based pricing. Central to this, he explained, is the "Agentforce work unit" a token spent executing a specific task rather than generating a general language response.
Traditional SaaS monetised human users and software licences. Agentic systems instead charge for actions, workflows and business outcomes. The commercial model follows the architecture so that if agents are the primary system users, pricing should reflect what agents do rather than how many humans hold seats.
Afshar also spoke about multi-agent orchestration, digital twins and ambient computing as the next wave of developments reinforcing this direction. These point towards the same direction of interfaces becoming optional, APIs becoming the primary point of access, and agents operating as autonomous system actors.
The bigger story may not simply be that Salesforce is adding AI capabilities to CRM. The company appears to be repositioning enterprise software itself around AI-native operation, where interfaces become secondary and agents become primary system users.
The Architecture is Already Changing
Vala Afshar’s interview did not introduce Salesforce's headless strategy for the first time, but it offered one of the clearest articulations yet of how the company sees enterprise software evolving in the AI era. As vendors redesign platforms for autonomous agents rather than solely for human users, the headless enterprise may become one of the defining architectural trends of the AI era.