Beyond Time-and-Materials: Pricing Software When Effort No Longer Predicts Output

Time-and-materials pricing has run enterprise software delivery for decades on one quiet assumption: that hours worked are a fair stand-in for value delivered. Buy more hours, get more software. Bill by the hour, and the invoice roughly tracks the work.

That assumption is now false. Agentic AI has broken the link between effort and output, and once that link breaks, pricing effort stops making sense. This is the first and most concrete way AI rewrites the software services contract, and it is worth understanding on its own before discussing what replaces it.

Time-and-materials was never really about time. It was a proxy. In a labor-driven model, output scaled with headcount and hours, so effort was the most auditable thing both sides could agree to measure. The buyer could count people and days. The provider could staff to demand. Neither had to agree on what a feature or a migration was worth, only on what it cost to produce.

That alignment is why the model lasted. It was honest about its own logic: both the buyer’s cost and the provider’s cost moved together, in the same direction, at the same rate.

The proxy only works while hours predict output. AI dismantles that at two levels.

First, human effort was already a poor measure of engineering value. Microsoft’s Time Warp study, a 2024 survey of 484 developers, found that writing code fills only about a tenth of the average developer’s week, behind meetings, communication, and coordination. Hours billed were always a blunt instrument, capturing time in seats rather than software shipped.

Second, agents sever the connection outright. When an agent library and a small team of engineers deliver what once took a large team, output no longer tracks the number of people or the hours they log. A provider can produce more, faster, with fewer billable hours. Under time-and-materials, that shows up as a smaller invoice for a better result, which is precisely backward from how the model is supposed to reward good work.

Effort and output have decoupled. Continuing to price the effort means charging for the wrong thing.

Here is the part that makes time-and-materials untenable rather than merely outdated.

Under an hourly model, a provider that adopts AI has a direct financial reason not to use it. Every hour AI removes is an hour that can no longer be billed. The more productive the provider becomes, the less it earns for the same outcome. The pricing model and the technology are pulling in opposite directions, and the buyer is the one who absorbs the misalignment.

No enterprise wants to pay a partner by a unit the partner is now motivated to inflate. Yet that is the position a time-and-materials contract creates the moment agentic delivery enters the picture. The model quietly rewards slower work at exactly the moment the technology makes faster work possible.

A buyer cannot fix this by negotiating a lower rate. The problem is not the price of the hour. It is that the hour is no longer the right thing to buy.

If effort is the wrong unit, the alternative is to price the result. Instead of buying a number of engineers for a number of months, the buyer contracts for a defined outcome at a defined cost, with payment tied to something the finance team already tracks: a delivery milestone, a cost of delivery reduced against a baseline, a release cadence, a quality threshold held.

This is the practical shape of Services-as-Software, delivery bought as a measurable result rather than as staffed capacity. It also realigns the incentive that time-and-materials broke. When the provider is paid for the outcome, using AI to reach it faster becomes an advantage rather than a revenue cut, and the buyer’s interest and the provider’s interest point the same way again.

The mechanism has a name in practice. Ascendion structures commercials around outcomes through what it calls AI Commercials, priced against the AI-driven productivity and savings delivered, in place of the heritage model of billing for time and materials. In production that choice has produced double-digit savings that free capital for reinvestment. It is the commercial expression of AI Arbitrage, using AI agents and AI-augmented teams to deliver engineering outcomes at lower cost and higher speed, and passing the economics to the client rather than absorbing them as billable hours.

The shift does not remove the buyer’s job. It changes it.

The question moves from what will this cost per person to what will you commit to deliver, and how is that measured. That requires a named, measurable outcome and a documented baseline at the start, because a result is only priceable if both sides agree what it is before the work begins. It also requires a provider that can actually control the outcome, which depends on a mature platform and the engineering depth to stand behind a commitment rather than a timesheet.

For the full framework, including where delivery risk moves and how to structure the contract, see our companion piece on outcomes-based commercial models. The narrower point here is simply this: the billable hour is no longer measuring the thing you are buying.