People
See each person's AI use: spend by week, models, the prompt cache, failed calls, their cap, and the coach's last note.
Every figure, and how it is computed
- Last 4 weeks
- Spend each week, against the person's usual week.
- Cache
- Prompt tokens read from the provider's cache, against your organization's.
- Cap
- Their budget in the gateway, and whether it refused them.
- Coach
- The weekly developer coach's last private note, and whether their figure moved the week after. Never the note's words.
What you do here
- Filter to people refused at their cap, or with a low cache.
- Open a person for their models and days.
- Raise a cap from their rail.