For twenty years, oncology practices have purchased software by the seat. Per-provider pricing. Annual contracts. Implementation fees. The SaaS model assumes that software is a tool humans use to do work.
AI agents break this assumption. When an agent can autonomously complete a prior authorization, schedule a follow-up, or reconcile a claim — the software isn’t a tool anymore. It’s labor. And labor isn’t priced per seat. It’s priced per task.
The Pricing Inversion
Consider the economics of prior authorization. A human employee costs $45,000–65,000 per year and can process approximately 15–20 authorizations per day. An AI agent running on cloud infrastructure can process thousands per day at a marginal cost approaching zero.
The SaaS vendor charging $500/month per provider for “AI-powered prior auth” is pricing based on the old model. The value-aligned price is per successful authorization. This changes everything about vendor selection, contract structure, and build-vs-buy decisions.
What This Means for Oncology Practices
Oncology practices should start evaluating every SaaS purchase through this lens: is this tool going to become an agent within 24 months? If yes, the current pricing model is a bridge. Don’t sign long contracts.
The practices that recognize this shift early will negotiate better terms, avoid lock-in, and position themselves to adopt agent-based architectures when they’re ready. The ones that don’t will be paying per-seat prices for per-task value.