Why per-seat fees are dead in the AI era.
When agents are users, not integrations, the old pricing model breaks.
Per-seat pricing made sense in a world where the only people doing work in a SaaS tool were people. You hired a salesperson. You bought them a CRM seat. You hired a support agent. You bought them a help-desk seat. Headcount and software cost grew together. The model was clean enough that nobody questioned it for two decades.
That world is ending. Once AI agents can run conversations, read records, and follow plain-language rules, the question of who counts as a “user” stops having a clean answer. Every per-seat invoice is a relic of an assumption that no longer holds.
This is an illustrative deep-dive into the model, not a price list. The figures below are examples chosen to show how the economics shift; current pricing lives on the pricing page.
Where per-seat pricing came from
Per-seat pricing solved a real problem. Software companies needed predictable revenue tied to a real input — and headcount was a real input. A customer with twenty salespeople had a bigger problem than a customer with two, and they got priced accordingly. Buyers accepted the model because hiring a salesperson and buying a seat were already the same decision in the budget meeting.
The model has hidden costs that compound as the team grows. Every new hire is a new license. Every team that wants to use the tool has to fight a procurement battle. Cross-functional adoption — the operations person who needs to read a deal, the engineer who wants to see a ticket — gets blocked by per-seat math. The tool that works for the sales org alone is not the tool that runs the business.
Why AI breaks it
When an AI agent can do the work of a person on a channel — answer the phone, reply to the chat, update the record — the per-seat model has a choice. Either the agent is “a user” and gets billed like a user, or the agent is not “a user” and the model has a hole in it.
Both options break the model.
If the agent is a user, the customer’s spend grows every time they automate something. The team that automates well pays the same as the team that does everything by hand, even though the work is being done by software the vendor barely touches. The incentive points away from automation.
If the agent is not a user, the customer pays for human seats while the agents do most of the work. The vendor’s revenue collapses as automation succeeds. The economic logic of per-seat pricing inverts.
The honest answer is that “seat” was always a proxy for value, and the proxy has stopped working. Conversations, call minutes, and automation runs are the things customers actually consume. Pricing should track those.
A worked example
The math below is illustrative. AYBIZA is launching, so these numbers are not customer averages — they are the public list pricing of category vendors compared against AYBIZA’s published Scale tier. Treat this as an order-of-magnitude argument, not a quote.
Imagine a 125-person company organized into five departments of about twenty-five people each — sales, support, and a few more. The legacy stack each department runs looks something like this on public list pricing:
- Voice AI vendor: $2,000 to $10,000 per month per department.
- CRM at twenty-five seats: $1,500 to $3,000 per month per department.
- Customer support tool: $300 to $1,500 per month per department.
- Team chat: $200 to $500 per month per department.
- Project tool: $200 to $1,000 per month per department.
- Scheduling tool: $100 to $300 per month per department.
That is roughly $4,300 to $16,300 per month per department at list price. Across five departments, the company is staring at $21,500 to $81,500 per month — most of which scales with seat count, not with the value each department extracts.
AYBIZA Scale is $997 per month for the full platform with unlimited users. Bring your own keys on AI, voice, telephony, storage, and web search and the per-credit cost drops up to 75 percent below the managed rate when your keys cover every layer; mixed setups typically save 33 to 50 percent. Credits pool across every department, and each one can bring its own keys.
The point is not the exact dollar amount. The point is that the per-seat tax that the legacy stack charges has nothing to do with the work being done. When agents do most of the work, paying for human seats is paying for the wrong thing.
What is next
The pricing model that replaces per-seat is not new. It is just unusual in B2B software. Pay for the work that gets done. Credits per conversation, per call minute, per automation run. The team can grow without a license per person. Automation makes the bill smaller, not bigger.
There will be three pricing tiers in this era — a free tier for solo operators and small teams, a recurring tier for predictable monthly volume, and a custom tier for enterprise scale. None of them are billed on heads.
Who benefits first
Growing companies benefit first. Per-seat fees multiply across every department added. A pricing model that does not scale with seats turns growth into a profit driver instead of a margin sinkhole.
Lean teams benefit second. The operations person who needs read access to a deal can have it without a license fight. The engineer who needs to see a ticket can see it. The full team uses the platform from day one.
Enterprises benefit third. Procurement teams who have spent years justifying seat sprawl get to retire the math. The replacement is volume rates and BYOK contracts that reuse the negotiations they already have.
The vendors that resist the shift will keep their per-seat invoices a little longer. The customers who switch will keep their margin.