The Billing Ladder: Five Ways to Price an AI Agent
AI agent pricing is best understood as a ladder — seats, tokens, conversations, resolutions, outcomes — where each rung shifts the cost of a failed attempt.
The "Billing Ladder" is a framework for understanding different pricing models for AI agents. It consists of five rungs, each representing a different way the vendor measures and charges for service. Rung 1 involves paying per seat, regardless of usage, while Rung 2 charges per token or minute, paying for consumption without regard for quality.
Rung 3 charges per conversation or session, paying for effort regardless of outcome. Rung 4 charges per resolution, only paying for completed work. Rung 5 charges per business outcome, paying for the result the work was supposed to produce.
Comparing pricing models on price alone without considering the underlying risk allocation is misleading, as they are denominated in different units. The key metric to consider is the effective cost per real resolution, which normalizes all pricing to the same standard. Most vendors operate on the lower rungs of the ladder, as outcome-based pricing only works where success rates are high.
Adoption of outcome-based pricing is still limited, as it requires high success rates and accurate attribution to profitability. The ladder's highest rung is the most difficult to reach, as it depends on the ability to defend measurements in disputes.
Written by urgent.news from HackerNoon's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
