Feature docs

AI usage dashboard on Vareons (/dashboard/ai-usage)

The /dashboard/ai-usage page is your store AI activity mirror: token summaries, per-prompt logs, 30-day CSV export, usage insights and distribution, and an audit table of the last 20 assistant tool executions.

Last verified: 2026-09-19

The usage page lives at /dashboard/ai-usage and requires dashboard login. UI access is tied to the /dashboard/analytics plan page entry, but every row stays scoped to your session tenant.

It shows token summary cards for the month or the rolling 30d window, a prompt-log table with all, errors, or high-usage filters, a 30-day CSV report export with the same days/filter semantics, an enterprise insights panel from GET /api/ai-copilot/usage/insights, token distribution bars from GET /api/ai-copilot/usage/distribution for seven days, and a tool audit table from GET /api/ai-copilot/usage/tool-audit capped at 20 executions with a days window aligned to the logs window.

The interface follows Arabic and English direction, locale number/date formats, and the global light/dark theme. It is the daily governance tool: you see what the assistant consumed, which prompts failed or consumed heavily, and which tools actually executed after your confirmation.

Setup steps in Vareons

  1. Open the usage page

    In the dashboard open /dashboard/ai-usage. Summary cards appear at the top with the default range; the page needs dashboard login only.

  2. Pick the range and review logs

    Switch the summary between month and 30d, and filter prompt logs across all, errors, or high usage. Keep the same days window when comparing with the tool audit table.

  3. Review insights, distribution, and audit

    Read the insights panel for recommendations, the distribution bars for where usage concentrates, and the audit table for the last 20 tool runs: which tool, when, and whether it succeeded.

  4. Export the report when needed

    Use Export report to download a 30-day CSV with the shown filter, and keep it for review, accounting, or sharing with a partner.

Frequently asked questions

What is the difference between month and 30d ranges?

Month is the calendar-month summary; 30d is a rolling last-30-days window computed from the prompt log rather than rollups.

How many executions does the audit table show?

Up to the last 20 assistant tool executions, with a days window aligned to the logs window in the 30-day view.

Can staff see the store usage?

The page sits in the protected dashboard and rows are scoped to the session tenant. Staff visibility follows your plan page policy and their permissions.

Does the export match the shown filter?

Yes. The CSV export uses the same days and filter semantics shown in the logs for the last 30 days.

Where do I add an API key from this page?

Through the Add API Key button, which takes you to /dashboard/ai-usage/api-keys to add, save, and test provider keys.

Troubleshooting

Numbers look off? 1) Compare the same range: a calendar month differs naturally from the rolling 30d window. 2) The errors filter shows failures only — switch to all before judging volume. 3) The audit table shows only the last 20 executions, not full history; use the export for the full picture. 4) If no data appears, confirm you are signed in with the right account for the right store.

Important limitations

• The audit table shows only the last 20 executions. • Export covers only the last 30 days. • Distribution is computed for seven days. • The page reports usage; it does not itself enforce an automatic cap.

Related links

تواصل معنا على واتساب