Professional services have historically billed for time, and software has billed for seats or usage, both proxies for value rather than value itself. When AI can complete a defined task end to end, resolve a support ticket, draft a compliant contract clause, qualify a sales lead, pricing can shift to the outcome directly: pay when the ticket is resolved, not for the hours spent trying.
This is a genuinely disruptive commercial shift, not just a billing tweak, because it changes who bears the risk. Under time-based billing, the client bears the risk that the work takes longer than expected. Under outcome pricing, the provider does, which only works if the provider is confident enough in the AI system’s consistency to absorb that risk at scale.
That confidence requirement is exactly why this model is spreading fastest where measurement is easiest and slowest where “the outcome” is ambiguous or hard to attribute cleanly to the AI’s work alone. The commercial model is downstream of whether you can actually prove the result, not the other way around.