Cloud and billing
Check deployment status, server activity, database storage and AI spend on the Cloud & Billing page, with shortcuts into the cloud console where billing lives.
Overview
Cloud & Billing answers "is the service healthy, and what is it costing?" in one place. It shows the current deployment, server activity, database storage against its quota, and AI spend, with links into the cloud console for the actual bill.
Before you start
- You must be a master admin. The page isn't visible to anyone else.
- Some figures only appear on the deployed service, not on a developer's local copy.
Steps
- In the sidebar, under Workspace, click Cloud & Billing.
- Google Cloud deployment shows Platform, Project, Region, Service, Revision and Scaling. If you see Not yet deployed to Google Cloud, the service hasn't been deployed there yet. Use Open Cloud Run console to jump to the service.
- Server activity shows instance activity, including Cold starts and Active time.
- Billable time is the only figure in this panel that maps to money.
- Idle time is free — when no one is using the app, it scales down and costs nothing.
- Database shows the database's Provider, Region and Connection, plus Storage used against the plan, Plan, Est. cost, Tables and Rows (approx). A link opens the database provider's console.
- AI spend covers the last few days of AI usage: Est. cost, AI calls, Tokens in / out, Failures and a By provider breakdown.
- Billing & cost — open in Google Cloud Console gives one-click links to Billing overview, Billing reports (cost), Budgets & alerts, Cloud Run services, Artifact Registry (images) and Build history.
Example
At the start of the month, the master admin opens Cloud & Billing. Server activity shows mostly idle time and a few cold starts, database storage is well under quota, and AI spend is almost entirely free-tier. They click Budgets & alerts to confirm the monthly alert is still set, and close the page.
FAQs
Why does AI spend here differ from AI Insights?
This page counts every underlying provider call — retries, multi-pass extraction and per-page scans — so it's closer to what providers bill. AI Insights counts one event per feature use, so its numbers are lower.
Why is the first page load sometimes slow?
When the app has been idle, it scales down to save money. The next visit starts it again, which shows up as a cold start.
Are the cost figures exact?
No. They are estimates. The cloud console and each provider's console are the source of truth.
Troubleshooting
"Couldn't load cloud status."
The page couldn't reach the status service. Refresh. If it persists, check that the service is running from the cloud console.
"Cloud Monitoring could not be read"
The server doesn't have permission to read monitoring data. Whoever manages the deployment needs to grant the service read access to monitoring. Instance metrics will appear once that's done.