AI for Deploy, Backup, and Remediation

Who this is for

Users who want practical examples of using AI Pilot for the three most common infrastructure workflows: deployments, backups, and problem remediation.

What you will complete

Learn the exact prompts and workflow for using AI Pilot to deploy a site, create a backup, and diagnose and fix a common infrastructure problem.

Before you begin

  • AI Pilot must be enabled with Write Actions on.
  • For deploy workflows: a site or app must exist and have a git repository connected.
  • For backup workflows: at least one server must be connected.
  • Owner or Admin role required for all write operations.

Using AI Pilot for deployments

Prompt examples

Trigger a deployment: "Deploy the latest version of my site named my-site-name on server-name."

Deploy a specific commit: "Deploy commit abc1234 to my-site-name on server-name."

Rollback a deployment: "Roll back my-site-name to the previous deployment."

Check deployment status: "What is the current deployment status of my-site-name?"

What to expect

  1. AI Pilot confirms the site name and server.
  2. It shows a plan: the deployment steps it will execute.
  3. An approval card appears — impact level is typically Caution.
  4. After you approve, the deployment starts and the Activity Center tracks progress.
  5. AI Pilot reports the result. If deployment fails, it shows the error and suggests next steps.

Tips

  • Always specify the site name and server name explicitly in your prompt. Avoid "my site" — use the actual name.
  • If the deployment fails, ask: "What went wrong with the last deployment of my-site-name?" The AI will read the logs and explain the error.

Using AI Pilot for backups

Prompt examples

Create an on-demand backup: "Create a backup of server-name right now."

Create a backup before a risky change: "Before I resize the disk on db-server, create a backup first."

Check when the last backup ran: "When was the last successful backup of server-name?"

Configure a backup schedule: "Set up a daily backup for server-name at 2 AM UTC."

What to expect

  1. AI Pilot confirms the target server.
  2. An approval card appears — impact level is typically Safe or Caution.
  3. After you approve, the backup starts. Backups for a typical server take 2 to 15 minutes depending on disk size.
  4. The Activity Center shows backup progress.
  5. AI Pilot confirms when the backup completes and shows the backup size.

Tips

  • For chained operations ("back up then resize disk"), AI Pilot will present the backup approval card first, wait for completion, then present the resize card.
  • If a backup fails, ask: "Why did the last backup of server-name fail?" The AI reads the error log and explains the cause.

Using AI Pilot for remediation

Remediation is the most powerful AI Pilot use case: diagnosing a live problem and fixing it.

Prompt examples

Diagnose a slow site: "My site my-site-name is responding slowly. What is causing it?"

Fix a service that is down: "The nginx service on server-name is not running. Please restart it."

Diagnose high CPU: "CPU on server-name has been above 90% for the past 30 minutes. What is using it and what should I do?"

Fix a failed SSL renewal: "SSL certificate renewal for my-site-name failed. Please investigate and try again."

Post-incident analysis: "The database server went down last night at 11 PM. What happened and how do we prevent it?"

What to expect

  1. AI Pilot reads the relevant metrics, logs, and alert history.
  2. It provides a diagnosis with the most likely cause.
  3. If a fix is possible, it proposes an action with an approval card.
  4. After you approve, it executes the fix and reports the result.
  5. If the fix works, the AI confirms. If it does not, it proposes the next step.

Tips

  • Include the time range in remediation prompts: "...in the past 2 hours" helps the AI focus on the right data.
  • After a fix, ask: "Please confirm server-name is healthy now." The AI will do a final read check.
  • For recurring problems, ask: "How can we prevent this from happening again?" The AI will suggest a monitoring rule or goal trigger.

What success looks like

  • Deployments: site is live on the new version. Activity Center shows "Completed." The AI confirms.
  • Backups: Activity Center shows backup completed with size and timestamp. The Backups page shows the new backup entry.
  • Remediation: the problem is resolved. AI Pilot confirms the service is running and the error is cleared.

Common errors and fixes

"AI Pilot cannot find my site or server by name" Cause: The name in your prompt may not match the exact name in the dashboard. Fix: Use the exact name as shown in the Servers or Sites page. Copy and paste to avoid typos.

"Deployment failed immediately after approval" Cause: The git repository credentials may have expired or the build step failed. Fix: Ask "What caused the deployment failure?" The AI will read the deploy log and explain.

"Backup started but shows Failed in the Activity Center" Cause: Disk space on the server may be insufficient for the backup snapshot, or SSH connectivity was interrupted. Fix: Check free disk space and server SSH status, then ask the AI to retry.

"AI Pilot diagnosed the problem but the fix approval card never appeared" Cause: The fix requires a permission level that is currently disabled (e.g., Service and SSH write permission is off). Fix: Check Settings → AI Agent → Agent Controls → Fine-tune Write Actions → Server Operations → Services and SSH.


Related articles