Seat-based pricing was the right model when every software interaction required a human user. The business logic was sound: more users means more value delivered, charge per user.
The logic breaks when AI agents start doing user-level work. When your customer's sales team uses an AI SDR that handles the top of funnel, that SDR isn't a seat. When their operations team deploys an AI analyst that runs weekly reporting, that analyst isn't a seat. The work is getting done. The value is being delivered. But the seat count is flat or shrinking.
This is the ticking time bomb: seat-based SaaS companies whose customers are deploying AI agents are watching their natural expansion motion stall. The customer is using more of your product's value, but they're not growing in a metric you're charging for.
The seat-based bomb has three detonation triggers:
Renewal negotiation based on actual usage. When a customer realizes their 100-seat license is being used by 40 humans and the rest are AI-assisted workflows, they'll ask why they're paying for 100 seats.
AI-driven headcount efficiency. Companies reducing headcount due to AI adoption are reducing seat counts proportionally. Your revenue follows their headcount down.
Agent interfaces that bypass seat tracking. If your product has an API and customers are using AI agents to access it, those API calls may not be tracked against any seat — creating usage without revenue.
The response requires a pricing model evolution, not just a price change. Move toward consumption (API calls, actions taken, outputs generated), outcomes (business metrics delivered), or value-metric seats that include agent actions. The specifics depend on your product, but the direction is clear: seat counting is the wrong unit for an AI-enabled world.
Defuse the bomb before the renewal conversation does it for you.