Usage Controls, Alerts & Recovery
Configure usage caps and alerts, diagnose paused runs, and restore blocked workspace capacity in Billing and usage. A notification threshold warns you, while a supported safeguard restricts new applicable work.
- Use it to
- Set resource hard caps and percentage alert thresholds, configure billing alert webhooks, and recover runs paused by exhausted balances or payment grace
- Find it
- Workspace menu → Billing and usage → Usage safeguards and notifications
- Limits
- Alert thresholds: 1% to 100% · Audit reason: 3 to 500 characters · Concurrent Step slots: 50 Free / 100 Paid
Managing budgets, caps, and alerts
Open the workspace menu, select Billing and usage, then open Usage safeguards and notifications from Usage or Billing. Check Applies to before editing. Only the current billing administrator can load private configuration and save changes.
| Setting | What it controls |
|---|---|
| Hard cap under a resource | The separate limit for that resource, where supported. Storage amounts use decimal GB. |
| Alerts on / Alerts off and Alert at | Whether that resource’s configured warning thresholds apply. |
| Also email | Extra email recipients for that resource’s alerts. |
| Also post these alerts to the webhook | Whether that resource uses the configured billing webhook. |
Prepaid Hard cap controls offer Zero, A separate limit, and Shared balance. Zero blocks new work for that resource. Shared balance removes the separate workspace limit without making the payer’s balance unlimited. Alert thresholds are warnings, not hard caps.
Lowering a supported safeguard below current usage blocks new applicable work. Already admitted work may finish. A storage cap does not delete existing files.
Available safeguards
An AI-token safeguard blocks new hosted AI work at its ceiling. A Retained-storage cap (GB) prevents new files from exceeding the cap while keeping existing files accessible. Warning thresholds (such as 80%, 90%, and 100%) create alert events without stopping work.
Save the controls
Enter your caps, unique alert thresholds (1 to 100 percent), and an Audit reason (3 to 500 characters). Then select Save controls. Saving controls and saving a billing webhook are separate actions. If a save is interrupted, use Recheck saved settings and compare your edits with Current saved settings before retrying.
Notification channels
Each metric has its own enablement, thresholds, additional email recipients, and webhook choice. These controls apply only to billing notifications, not automation execution-failure emails.
1. Email notifications
Add additional email recipients for any metric that needs extra usage warnings. Saving an address configures the rule; it does not confirm inbox delivery.
2. Billing webhooks
Configure the webhook endpoint
Check Applies to and enter the Public HTTPS URL and required reason. Then select Create webhook and signing secret or Replace destination and rotate secret.
Store the one-time signing secret
Copy the masked one-time signing secret into secure storage before closing the dialog. Verify the X-Glow-Signature header at your receiver.
Enable metric webhook alerts
Turn on Also post these alerts to the webhook for the desired metrics and select Save controls.
A lost webhook signing secret cannot be retrieved after the dialog closes. If lost, rotate the secret deliberately and update your receiver. Never paste secrets into tickets or screenshots.
Understanding pause states
When automation execution stops or waits, check resource balances, payment status, and execution capacity separately:
Steps requiring an exhausted balance pause with out_of_credits or
glow_ai_token_limit. Unrelated work can continue if its own allowances and
safeguards permit it.
An unpaid renewal can place the workspace in payment grace. Resolve every outstanding invoice before the deadline in Billing.
2. Payment Grace PeriodWaiting for execution capacity
Standard Free capacity is up to 50 concurrently claimed Step jobs and Paid up to 100. Lower operator safeguards can apply. Where a payer is shared, its active funded workspaces take turns within one aggregate ceiling and can borrow idle capacity.
Waiting consumes no retry attempt or additional allowance. After a plan change, new jobs are admitted as active execution slots complete. Contact support if queued runs remain waiting longer than expected.
Restoring service and recovering automations
Have the current billing administrator resolve funding, payment, or safeguard changes for the affected workspace. For sponsored workspaces, coordinate with the sponsor’s payer administrator.
Admin: restore allowance or resolve the invoice
If an allowance is exhausted, open Billing and usage → Usage to increase capacity or wait for the scheduled reset. Creating another workspace does not grant another Free allowance. An unfunded additional or replacement workspace needs a prepaid subscription; see Plans & Feature Comparison.
If an invoice is outstanding, open Billing and usage → Billing and use Pay outstanding invoice to resolve all open invoices.
Admin: confirm available capacity and billing status
Paid access depends on payment verification. Provider delays can keep execution blocked. Confirm the current Billing status and verify that balances and safeguards permit new work. Paying an invoice does not reset measured usage or automatically resume a stopped run.
Operator: inspect the stopped run before retrying
Check the affected run’s Execution Log, the Step’s Executions tab, and the destination app before retrying. A stopped run may already have sent a message or created a record in an earlier Step. See Testing & Debugging.
Operator: verify recovery with safe work
Confirm Live status only for automations meant to run automatically. Test webhooks with a fresh event against a safe destination, and check Next runs on scheduled triggers. Compare execution history with external sender logs before reprocessing any events missed during an exhaustion window.
- Run
- Open Billing and usage, select Usage safeguards and notifications, set an alert threshold with an audit reason, and click Save controls
- You should see
- Glow saves the threshold immediately and sends alerts when workspace consumption crosses that percentage.
What’s Next?
- Activity & Automation Monitoring: Check runs and alerts after restoring service.
- Testing & Debugging: Inspect results before retrying work.
- Step Credits, Tokens & Storage: Look up consumption and allowance rules.