Scry prices search by congestion, not by seat

Most software APIs charge by the call, by usage tier, or by the number of people using them. Scry is taking a different route with its programmable search API. Its pricing rises when demand for a particular type of query spikes, rather than staying flat for every request or charging per seat. You can see the company’s positioning on Scry’s website.

For an owner, the interesting part is not search itself. It is what this says about the economics of AI agents. Agents can create uneven demand. A workflow might run quietly most of the day, then launch hundreds or thousands of searches when a sales list is refreshed, an incident occurs, or a batch process starts. Flat pricing can make that expensive capacity hard for providers to plan, while fixed seat pricing makes little sense when software, not people, is doing much of the work.

Congestion-based pricing shifts some of that cost to the moments when a shared service is under pressure. That could help API providers manage unpredictable agent load. It also means buyers will need to understand when their automated workflows run and what query patterns they create. A cheap-looking API may cost more during busy periods, so usage controls and monitoring matter.

This is the kind of thing we build into an AI command centre, where you can see what automated work is running, where demand is building, and what it is costing the business.