Something structural is changing in how AI products charge for their value. The seat-based subscription — the model that defined SaaS for two decades — is losing ground to something older and simpler: pay for the outcome, not the access.

The examples are becoming visible. A customer support AI charges per resolved ticket. A contract analysis tool charges per document package. A proposal generator charges per draft. In each case, the pricing unit is a completed piece of work, not a user license or a monthly seat.

This isn’t a novelty. It’s a reversion. Before SaaS, professional services always charged by the deliverable. You paid the accountant for the tax return, not for access to the accountant’s knowledge. SaaS disrupted that model by making access cheap to distribute — once you’d built the software, adding another user was nearly free, so charging by user made sense. The economics have shifted again.

When an AI agent does the work, the marginal cost structure changes. The cost is no longer server infrastructure amortized over user count — it’s compute per task. The natural pricing unit follows the cost unit. And the natural pricing unit for compute per task is the task.

There’s a product philosophy embedded in this choice. When you charge per resolution, every decision your team makes gets evaluated through the lens of “does this make resolutions happen more reliably and efficiently?” You can’t optimize for engagement or session length or feature adoption in the traditional SaaS sense. You optimize for completed work. That’s a clarifying constraint.

It also changes the sales conversation. A per-seat subscription gets evaluated against budget. A per-resolution tool gets evaluated against what the resolution is worth. If resolving a commercial lease abstraction takes four hours manually and you can do it in fifteen minutes for twenty-five dollars, the comparison isn’t “is twenty-five dollars within budget?” — it’s “is four hours of analyst time worth more than twenty-five dollars?” That’s almost always an easy yes.

The resistance to this model comes from unpredictability. Buyers used to fixed SaaS costs don’t want variable bills that scale with usage. The answer is the hybrid: a base subscription that covers a predictable volume of work, with per-unit pricing for overages. This preserves budget predictability while keeping the pricing logic aligned with value delivered.

The per-resolution shift is still early. Most AI products are still charging seat-based subscriptions because that’s the model everyone knows how to buy. But the products that are growing fastest in 2026 are the ones that figured out how to price the outcome. +++