The Mirage of Outcome Pricing for AI Agents

AI agents are eating the web, as I’ve explained in my piece: The Agentic Web.

Yet, there is another reality: AI agents will also become the key software player in the coming decade. Meaning they’re swallowing up SaaS to turn it into something quite different.

Where SaaS has been, by definition, vertical, narrow, and based on enabling productivity. AI agents, as a software paradigm, will be horizontal, end-to-end, and based on allowing outcomes to.

From there, the vision is compelling: AI agents that charge based on the value they create, not the resources they consume.

Easy, isn’t it? Well, for now, not really!

In theory, let’s say you pay $15,000 for an AI agent that helps you ship a product 50% faster, saving $100,000 in engineering costs.

Everyone wins—providers capture fair value, users see clear ROI, and the ecosystem becomes sustainable.

But here's the uncomfortable truth: pure outcome-based pricing for AI agents is a mirage in the short term.

Not because the concept is flawed, but because the infrastructure—technical, organizational, and cultural—required to make it work doesn't exist yet.

And that is a real business transformation challenge that spans the technical and organizational requirements to make that happen.