Can old SaaS pricing survive a product category it wasn't built for? Not really.
Walk into almost any pricing page in 2026 and you'll find the same three ingredients as 2015: a per-seat fee, a handful of tiers named after metals or planets, and an "Enterprise — contact us" box at the end. Same old SaaS, same old tricks.
The assumption underneath all of it
Per-seat pricing rests on one assumption: value scales with headcount. More people logged in, more value being extracted, more you can charge. That assumption held for two decades because software's job was to make humans faster, not to replace the human step entirely.
Agentic products don't make a human faster at a task. They remove the human from the task. The pricing model built for "more people using this" has nothing to say about "nobody's using this, it's just running."
Three vendors, three patches, none of them the fix
Hyundai, Google, and Cadbury all rebuilt product lines around a shift they saw coming rather than patching the old one — that's the instinct missing from most pricing pages today. Instead, most SaaS vendors are patching:
Patch one: rename the seat. Call it a "workspace" or a "workflow" instead of a "user," keep the per-unit math identical. Doesn't solve anything, just relabels the invoice.
Patch two: bolt on usage credits. Keep the seat fee, add a credit meter for agent actions on top. Customers now pay twice for the same capability — once to access it, once to use it — and notice.
Patch three: quietly stop selling seats to agent-heavy customers and hope nobody asks why the pricing page doesn't mention it. This is more common than vendors admit.
The back door and the front door
There's a back door and a front door to this problem. The back door is what most companies are taking: keep the old model, add exceptions, hope the agent stuff stays a small enough percentage of revenue that nobody has to rebuild the pricing page. That works right up until it doesn't.
The front door is building pricing around what the agent actually produces — tickets closed, leads qualified, dollars recovered — and treating that as the sellable unit from day one. Harder to build. Doesn't require a rename in eighteen months when the back door runs out of hallway.
Less tricks, more back to basics: charge for what the product actually does, not for how many humans are technically allowed to watch it happen.
Both approaches still let your customer endlessly scroll through their dashboard wondering what they're paying for. Only one of them has a good answer.