AI agents are becoming the primary users of enterprise software, and that structural shift breaks most assumptions SaaS was built on. Seema Amble, Steven Sinofsky, and Elena Burger at a16z dig into what they call 'headless' software: applications stripped of their UI and accessed purely through APIs and agentic workflows. Salesforce's Headless 360 announcement is the clearest public signal that incumbents see this coming.
The conversation is worth reading in full because the most interesting argument is not the obvious one. SAP and Salesforce are not dying. Their stickiness comes from data gravity, compliance entanglement, and exception handling, not from their interfaces. The panel argues that 'vibe coding' your way into enterprise fails precisely because exception handling is the entire game, not an edge case. Productivity gains from AI create new workflows and new volume, not fewer jobs or fewer software seats.
The sharpest opportunity the panel identifies is in the layer between agents and legacy systems of record: the connective tissue, the MCP implementations, the workflow orchestration that enterprises cannot buy off the shelf yet. Startups that understand enterprise data architecture will win here. Founders who think replacing SAP is the pitch will not.
[WATCH ON YOUTUBE →]