AI Tool Cost is the monthly estimated cost per connected AI tool, plus a utilization view of how much of that paid spend is actually being used. It's the spend-and-utilization side of AI Governance — pair it with adoption and impact to evaluate AI return on investment.
What it measures
For each connected AI tool, the estimated monthly cost based on the data sources Leanmote can access, alongside utilization (the share of paid seats that recorded any activity in the period).
How Leanmote calculates it
GitHub Copilot — cost is pulled directly from the Copilot billing API: number of seats × seat price for each Copilot plan tier (Individual, Pro, Team, Business, Enterprise).
Claude Code — cost is estimated as active users × seat price for the period.
Cursor — cost is estimated as active users × seat price for the period.
"Active users" for Claude Code and Cursor is the count of distinct people who recorded at least one AI tool activity in the period (any telemetry event in ai_facts).
Utilization for Copilot is active billing seats ÷ provisioned seats from the Copilot billing API. For Claude Code and Cursor, there's no separate provisioned-seat concept today, so utilization is reported as 100% — the count of active users is the seat count we use for cost.
Default seat rates and overrides
Default per-seat prices today:
Copilot — taken from the Copilot billing API per plan tier.
Claude Code — $25.00 per active user, per period.
Cursor — $20.00 per active user, per period.
Claude Code and Cursor seat prices can be overridden at query time when calling the AI Governance API directly, but there is no admin UI to change the defaults inside the workspace today. If your contract differs (volume discounts, committed spend, custom seat prices), contact your account team and we'll update the configured rate.
How to interpret it
Cost rising with adoption — expected. Pair with AI Intensity and AI Lead Time Comparison to validate that the additional cost is producing gains.
Low Copilot utilization — paid seats are sitting idle. Reclaim and reassign before the next renewal.
Claude Code or Cursor cost growing fast — because cost scales linearly with active users for these tools, headcount changes show up immediately. Watch month-over-month trends with that in mind.
Cost flat with rising adoption — your seat plan is being maximized. Negotiate a better rate at next renewal.
Cost accuracy notes
Copilot is the most accurate — it's reading directly from the billing API. If billing access isn't granted, the panel shows what it can.
Claude Code and Cursor are estimates. Actual invoices may differ if your contract has bulk-license discounts, committed spend, custom rates, or a different billing cadence.
Bulk discounts and committed spend agreements aren't auto-detected.
What to do about it
Track cost per merged PR (Cost ÷ PRs merged) as a unit-economics indicator.
Investigate Idle users with paid seats — see AI User Engagement Segments.
Use trends quarterly, not monthly — usage and cost both swing with project intensity.
Related metrics
AI User Engagement Segments
AI Intensity
AI Governance metrics overview