Model Usage Visibility
See how your team uses AI. Understand the usage and cost.
See recorded model usage across user support, investigations, ticket assessment, and workflows. Review consumption by source, model, and day, with cost estimates based on configured pricing. Use that visibility to understand how AI contributes to your support work.
| Source | Prompt | Completion | Total | Requests |
|---|---|---|---|---|
| Investigations | 11,616,000 | 384,000 | 12,000,000 | 1,200 |
| End-User Conversations | 8,712,000 | 288,000 | 9,000,000 | 1,800 |
| Technician Conversations | 7,260,000 | 240,000 | 7,500,000 | 750 |
| Workflow Runs | 4,065,600 | 134,400 | 4,200,000 | 3,150 |
| Tickets | 4,356,000 | 144,000 | 4,500,000 | 900 |
| Executive Summaries | 290,400 | 9,600 | 300,000 | 30 |
Sample dashboard for a 250-endpoint MSP at the "Typical" usage profile (~150K tokens / endpoint / month). The real view inside your tenant streams live, attributable to the same source categories, with per-model + per-day drill-downs.
Why teams ask for this first.
Every team evaluating automated IT support worries about the same thing: unpredictable usage and surprise costs. The dashboard answers the three questions a CFO actually asks before signing.
Every token attributed to one of six sources: end-user chat, technician chat, investigation, workflow run, ticket, executive summary. Hover any tile to see the requests that drove it. No black-box billing.
Monthly estimate, daily run-rate, and per-endpoint average refresh live. Project against 250, 1000, or 10,000 endpoints with one input. Use it to lock annual budgets - or to compare against your BYO provider's bill.
Daily trend + tokens-per-request panel surfaces unusual usage the day it happens. Set alert thresholds per source or per model. New playbook spiking your investigations? You'll know on day one, not on the invoice.
See your real numbers in 30 days.
During the 30-day trial, review actual support activity and model usage in your own environment. The dashboard shown above uses illustrative data; it is not a live view of a customer tenant.