What changed
Legora has introduced a consumption-based pricing model for Agent Pro while keeping its broader platform commercial model intact. The company emphasizes consumption credits, real-time usage visibility and spend controls rather than a simple per-seat price for the agent workload.
Why it matters
AI agents can have materially higher and less predictable variable compute costs than traditional SaaS features. Separating agent usage from seats gives vendors a way to align revenue with cost, but it also shifts budgeting risk toward customers unless usage measurement and controls are clear.
Agent economics are being unbundled from seats
Legora's public materials describe Agent Pro as consumption-priced, while its broader platform remains sold through commercial contracts rather than a fully public self-serve price card. The distinction matters because customers can add AI workload cost without adding human users.
Controls become part of the product
Legora highlights real-time usage visibility and spend controls. In a consumption model, those controls are not admin polish: they are part of whether customers can safely deploy an agent across teams and matters without surprise bills.
Public unit economics are still incomplete
Legora does not publish a universal public conversion from credits to a standard customer cost per task. That means outsiders cannot responsibly calculate a representative effective price per legal workflow from public materials alone.
The broader lesson is architectural, not vendor-specific
For AI SaaS builders, the case illustrates a hybrid pattern: retain predictable platform or seat economics where they fit, meter expensive agentic workloads separately, and give customers tooling to understand and constrain usage.