For twenty years, buying software followed a simple script. A company picked a tool and paid a fee per user. The contract renewed a year later.
That script is starting to break down. AI is changing the unit of value software gets priced on.
When an AI agent can do the work, charging by the login stops making sense.
The Seat-Based Model Was Built for a Different Era
Per-seat pricing worked because headcount was a reasonable proxy for value. More employees on a CRM meant more deals tracked. It meant more work done.
That logic assumed a person had to sit at the keyboard. AI features now let one user trigger far more output than they could by hand. Usage-based pricing has moved from a niche experiment toward the mainstream as a result.
It has not fully displaced the seat, though. A 2026 analysis of usage-based pricing models found that only around 15% of software companies run a largely usage-based model. Most are still blending usage into an existing subscription.
Most vendors are not ripping up their pricing pages overnight. They are layering new units of consumption on top of the seat, testing which ones customers actually accept. That hybrid phase is likely to last for years, not months.
Why AI Agents Break the Per-Seat Logic
Picture one employee with a single CRM licence. Under the old model, that licence represents one person’s worth of work.
Give that employee an AI agent inside the same CRM. The licence stays the same, but it can now update records, qualify leads, and draft follow-ups all day. The seat count stays flat while the workload climbs.
Vendors are responding by testing several units of value at once. Some charge per task or per resolution. Others layer a flat “agent” licence on top of an existing plan.
Salesforce’s Agentforce is a visible example of this in practice. According to recent SaaS pricing data, it generates close to $800 million in annualised revenue. It runs several pricing structures side by side, from per-conversation charges to a flat agentic licence.
From Licence Management to Usage Management
Traditional SaaS spend management has centred on a narrow set of questions. How many licences do we own? Who is actually using them?
Those questions do not disappear in an AI-driven world. But they stop being enough on their own. Finance teams now also need to know which AI features drive cost.
Two companies with identical seat counts can carry very different bills. It depends on how heavily their teams lean on AI.
Which workflows are consuming the most resources? What does a completed task actually cost to run? These are new questions, and most spend management processes were not built to answer them.
Why AI Makes SaaS Costs Harder to Forecast
Forecasting a traditional SaaS bill used to be arithmetic. A hundred users at fifty dollars a month gave finance a stable number.
AI-linked pricing adds several moving parts at once: a base subscription, metered usage, API calls, and inference costs. These can all shift month to month. Many finance teams are now being surprised by their own bills.
Research on IT spending found that 78% of IT leaders had encountered unexpected charges tied to consumption or AI features. That points to a visibility problem as much as a cost problem.
Budgeting around a moving target is hard. Finance teams are used to locking in a number and revisiting it once a year. AI-linked spend does not sit still that long.
SaaS Spend Management Is Becoming AI Spend Management
It is the structural shift that counts now. The capabilities of AI are now embedded directly in products such as Salesforce, Microsoft 365 and HubSpot. Organisations aren’t even buying an ‘AI product.’
That raises a question most finance functions have not answered yet. Where does AI expenditure actually sit? It is no longer cleanly separable from the SaaS line item it lives inside.
So the cost can go unmonitored. It often takes a renewal or an invoice to force the conversation.
What This Means for CFOs and Software Buyers
None of this means finance teams need to overhaul their approach overnight. It does mean spend management needs to extend beyond licences and into activity.
A few practical shifts are worth making as this transition plays out.
- Track AI feature usage alongside seat counts, not as an afterthought to them.
- Ask vendors directly whether usage- or outcome-based pricing options exist before renewal.
- Build a wider margin into software budgets where AI features are involved.
- Revisit renewal and procurement calendars regularly, since pricing structures are still changing fast.
For teams auditing existing software spend, SaaS spend management tactics can provide a useful framework for identifying unnecessary costs and unused tools.
The Road Ahead
Pricing models across the SaaS industry are likely to keep fragmenting. Seat-based, usage-based, outcome-based and hybrid approaches are all being tested in parallel. Often, it’s by the same vendor for the same product.
One thing is already clear. The companies that manage software spend well will be the ones tracking activity and outcomes, not just logins. That is a meaningfully different discipline than the one SaaS spend management was built around.
The seat will not disappear completely. But it is no longer the whole story, and treating it that way is how budgets get blindsided.
