AI Usage & Spend
Track AI coding tool spend, adoption, and usage per user, team, and model.
What AI Usage & Spend shows
AI Usage & Spend brings the cost and adoption of your AI coding tools into one place: how much you are spending, who is active, which models and products get used, and how that changes over time. It reads from the AI tools you connected (Cursor and Claude), so it fills in once at least one of those is set up.
There are four views, reachable from the Insights area:
- By User: spend and activity per person, with top spenders and a per-member detail page.
- By Team: the same rolled up to teams (populated from your org chart), with a per-team detail page that mirrors the member view.
- Spend: workspace totals, trends, and the provider / model split.
- Costs by LLM Model: spend and token volume grouped by model.
Cursor auto-routed usage is labeled Auto on model lists, the provider / model split, and the daily spend chart.
Supported providers and features
Not every provider exposes the same data. What you see depends on the tool and, for Claude, on your Anthropic plan. This matrix reflects what the platform can show today.
| Feature | Cursor | Claude Team | Claude Enterprise |
|---|---|---|---|
| Total spend | ✓ | ✓ | ✓ |
| Spend by user | ✓ | Limited | ✓ |
| Spend by team | ✓ | Limited | ✓ |
| Spend by model | ✓ | — | ✓ |
| Spend by product 1 | — | — | ✓ |
| Token usage | ✓ | — | ✓ |
| Sessions / events | ✓ | — | — 2 |
| Daily activity calendar | ✓ | Limited | ✓ |
| Daily spend & usage trend | ✓ | — | ✓ |
| Time presets + custom range | ✓ | ✓ | ✓ |
Legend: ✓ supported · Limited partial data from the provider · — not available
1 Product split is unique to Claude Enterprise (for example Claude Code, Chat, Cowork). Cursor has no product concept, and Claude Team does not expose it.
2 Claude’s cost report has no session concept, so Enterprise members show an activity cadence instead: active days (or active months for older, month-stamped data) and average spend per active day.
On the Team plan, Anthropic exposes workspace-level usage but little per-person detail. Per-user and per-team spend may be partial. Claude Enterprise provides the full breakdown.
Time range
Every AI Usage & Spend view shares one time filter, granular to the day:
- Presets: Last 7, 30, 90, or 180 days.
- Custom range: open the calendar and pick a start and end date to focus on a specific slice. You can go back up to 180 days, and future dates are not selectable.
The preset dropdown and the calendar stay in sync: choosing a preset moves the calendar to that window, and editing the calendar switches the filter to that exact custom range. Dates follow your workspace timezone from Settings → Preferences, so day boundaries match the dates you see elsewhere.
Member detail
Click any person in the By User list to open their detail page. It carries the same time filter, so you can re-scope the range without going back to the list. It includes:
- KPI cards: spend, tokens, and an activity metric (sessions for Cursor, active days for Claude Enterprise) with the matching average.
- vs team average: how the member compares to their team, shown as an increasing or decreasing pattern rather than a flat number.
- Daily spend & usage trend: a per-day chart of spend split by product or model, with a token-volume line. This is most useful for Claude Enterprise now that its usage syncs daily.
- Product usage and Model usage: where the spend went, by Anthropic product and by model.
- Activity calendar: a heatmap of active days across the selected period. Toggle it to shade by spend or by tokens, and hover any day to see that day’s spend and token count.
Team detail
Click any team in the By Team list to open its detail page. It gives teams the same depth as the member view, aggregated across everyone the org chart attributes to that team:
- KPI cards: total spend, members in scope, active members, and average spend per active member.
- Spend by member: a ranked breakdown of who spent what within the team — click a member to open their own detail page.
- Product usage and Model usage: the team’s spend by Anthropic product and by model.
- Daily spend & usage trend: a per-day chart of the team’s combined spend split by product or model, with a token-volume line.
- Activity calendar: a heatmap where a day counts as active when any team member had synced usage, shaded by the team’s total spend that day.
Data requirements
| View | Required integration |
|---|---|
| Any spend or usage view | Cursor and/or Claude (see Integrations) |
| Per-product breakdown | Claude Enterprise |
| Team roll-ups | HR org chart (CSV), Personio, or ZenHR |
Freshness follows the platform’s normal schedule: first numbers land within a few hours of connecting, and dashboards refresh automatically about every five minutes while the tab is open. See the Cursor and Claude pages for connector specifics.
When the usage pipeline is incomplete
Those same screens, including member and team detail, keep their charts if AI usage ingest is still catching up. One banner can appear on the overview, By User, By Team, Spend, Costs by LLM Model, and the detail pages when that pipeline is Backfilling, Not started, still refreshing, Failed, or the window is stale or missing days.
A Backfilling banner shows a progress percent when that number exists (from 0 to 100). A refresh with no percent reads as still refreshing, with no countdown. Failed, stale, and missing days link to Data coverage filtered to AI usage (/workspace/data-coverage?pipeline=ai_usage). The banner stays until the pipeline is current again. It is absent when the pipeline is healthy, when Data coverage has no AI usage pipeline, and when the coverage request is loading or failed. Any role that can open the page sees it.