Enterprise AI Measurement Guide
Agent Effectiveness
Spend
How does the amount we're billed for AI coding agents compare to what our telemetry shows we actually consumed, and what does the gap tell us about our agent cost governance?
What it shows
The Agent Effectiveness Spend card shows AI coding agent spend with a toggle between two cost bases: Billed (invoiced costs from vendor statements) and Observed (Larridin's usage-derived cost estimate based on behavioral telemetry). The trend shows how agent spend is moving week over week. This is the spend view scoped specifically to agent activity, distinct from the total AI Spend Intelligence view, which includes all AI tool categories.
Why it matters
The gap between Billed and Observed agent spend surfaces in two scenarios that have opposite implications. If Observed exceeds Billed, the organization is consuming more than its flat-rate licenses cover: usage is running ahead of what the invoices show. If Billed exceeds Observed, the organization may be paying for capacity it is not using, or the vendor billing model includes overhead the usage data does not capture. Either gap is a budget management signal that only appears when both numbers exist in the same view.
The Larridin angle
Scoping the Observed/Billed comparison to agent activity specifically, rather than all AI spend, allows engineering leaders to isolate the cost of agentic workflows from the cost of inline AI assistance, which have fundamentally different consumption patterns and different governance implications.