Know why a backup failed, and what to try first.
When a report says a backup failed or warned, or no report arrives, BackupSentinel works out the likely cause from the report email and the backup's history, and lists the fixes to try, most likely first.
It reads what the backup software wrote.
The lines of the report email that describe the problem, the product that sent it, the backup’s last ten results, and how many other backups show the same error right now. A single failure between good runs reads as something a retry clears; the same error night after night reads as storage, credentials or configuration.
It knows the usual failures of Veeam, Acronis, Synology Active Backup and Hyper Backup, Datto, Proxmox, NAKIVO, MSP360, Cove, Duplicati and Microsoft 365 backups, and Windows error codes such as 0x80070070.
Before anything leaves BackupSentinel
- Host names, IP addresses, email addresses, file and share paths, licence keys and ids are replaced with placeholders.
- Client, backup and workspace names are removed too.
- The redacted text goes to Anthropic, which runs the Claude model. Anthropic does not train its models on it.
- Every explanation and the exact redacted text sent for it are kept, so an owner can see what was shared.
Subprocessors lists Anthropic with the other services we use.
Where you see it.
Explanations are worked out in the background within about ten minutes of the alert, so alerts are never slower. On any open problem you can also ask for one with Explain this failure.
| Where | What it shows |
|---|---|
| Needs attention | Under each failed, missing or warning problem: the likely cause in one line, opening to the ranked fixes. |
| The backup's page | Under Why it needs attention, opened, with how sure the explanation is and whether a retry is likely to work. |
| Daily digest | A What to try line for each open problem that has an explanation. |
| Alert emails | When the same backup raises the same alert again, the email carries last time's likely cause and first fix. A new error is never waited for. |
| Weekly summary | The same cause on several backups in the week, often one NAS, one repository or one credential. |
| MCP server | The explain_failure tool, so your assistant can answer why a backup failed. |
On every plan.
Starter includes 10 new explanations a month. Growth, Pro and Enterprise have no monthly count. The same error on the same workspace is explained once and reused, so a backup that fails the same way every night does not use your allowance again.
Explanations are written in English, French, German, Dutch, Spanish, Italian or Portuguese: the language of the report email, or the one set in Settings → Billing.
What it does not do.
- It suggests; it does not diagnose. Each explanation says how sure it is, and a short error line gets a low-confidence answer that starts with the check that tells the causes apart.
- It changes nothing: it does not retry jobs, log in to devices or touch your backup software.
- It works from the report email. A run reported through the API has no email text, so it is explained from the backup's history only.
- A problem that closes on its own (the next report is OK) closes as before, whatever the explanation said.
See your own failures explained.
Start the trial, add a client and forward one report email.