Why is software priced per seat?
Software is priced per seat because it was built for humans to operate. A seat measures access: one person can log in and use the tool. The value delivered depends on how much that person does with it, which the vendor cannot see, so the vendor charges for the access and the buyer bears the productivity risk.
This model breaks when the software does the work itself. An AI employee that reconciles a ledger or drafts a shortlist does not need a seat; it needs a budget for the work it will do.
What does cost per outcome mean?
Cost per outcome means paying for the work produced rather than for the right to produce it. On ftes.ai a plan buys a block of work metered in tokens, and every Digital FTE the organization hires draws from the same wallet. Drafting, fact-checking, measuring, briefing, and proposing consume tokens; reading the workspace does not.
The unit is the work, which makes the purchase comparable to capacity rather than to licences. Hiring a second or third Digital FTE does not add a seat charge; it adds work drawn from the same block.
How does this change comparison?
Comparison shifts from seats to outcomes. Instead of asking what a licence costs per user per month, the question becomes what a reconciled month, a published and cited page, or a ranked shortlist costs, and how that compares with the cost of producing the same outcome another way.
The workspace shows usage per Digital FTE, so the cost of each role's work is visible rather than blended into a licence fee. That visibility is what makes the outcome comparison possible.
- Seat model: price × users, value depends on usage you cannot see
- Outcome model: price × work, usage visible per role in the workspace
- Adding a role: seat model adds licences; outcome model adds work from the same wallet
- Overspend risk: seat model unlimited usage; outcome model capped by the wallet
What about spend control?
Metered pricing raises a fair question about runaway spend, and the answer is the wallet. When the block is exhausted, the Digital FTEs pause and say so in the workspace; nothing runs on credit. Top up or wait for the next cycle and they resume where they stopped.
Inside the block, budgets can be set per FTE and per mission, and spend against budget is visible in the morning brief. Spend control is a feature of the model rather than a policy layered over it.
How should a buyer use this?
Use it by working out the outcomes you need per month and what they cost today, then comparing with the metered cost after a pilot. A finance team that knows what its close costs in hours can compare that with the tokens a Finance Controller Digital FTE consumed to prepare it.
The comparison is honest only if the baseline is recorded before the pilot and the outcome is measured the same way afterwards. Cost per outcome is a better unit than cost per seat, but it still needs a real numerator and a real denominator.
What is a labelled example of the comparison?
As a labelled example with your own figures to fill in: a seat-priced tool for a support team costs a licence per agent per month, and the outcome it contributes to, tickets resolved, depends entirely on how many hours the agents spend using it, which the tool's invoice does not show. A Customer Support Digital FTE costs the tokens it consumed in the month, visible in the workspace as usage for that role, and the outcome it produced, first-line tickets resolved with a cited source, is recorded in its Targets tab with the evidence per item.
Dividing the second by the first gives a cost per resolved ticket that can be set against the baseline cost of resolving the same tickets by hand, recorded before the pilot from the agents' own time. The seat-priced tool cannot produce that number, because its cost is attached to people rather than to work. The comparison is only honest with the baseline recorded first and the outcomes measured the same way afterwards, but once it is done, it is the comparison a finance leader can actually use, and it is the reason the pricing unit on ftes.ai is the work.