Attributes model spend to tenants, features and tasks, with budgets and alerting.
Tags every call so spend can be attributed and controlled. Necessary because inference cost scales with usage while subscription revenue does not, so a heavy user can become individually unprofitable with nothing appearing wrong in aggregate.
Multi-tenant products, and any deployment where usage is not proportional to revenue.
Fixed-volume internal use with a predictable ceiling.
Budget enforcement must degrade gracefully rather than failing hard. Cutting a tenant off mid-close because a monthly budget was exhausted is worse than the overspend.
Attribution by call count rather than token consumption, which understates document-heavy tenants by a wide margin.
A tenant whose evidence-validation usage triples after adopting the feature, visible as cost per tenant per month before it appears on an aggregate bill.
Two different absences share this shape. Foundational solutions get built whatever the domain, so no domain links them; the rest are solutions this domain genuinely does not reach for. v_ai_solutions_unlinked separates the two.