Manage application installations with Copilot

Step 1: Pull the list of affected devices and the specific errorCopilot

Export the failed-deployment list from your management console with the device name and the exact error code for each — a prompt built on "installs aren't working" with no specifics produces generic troubleshooting advice that doesn't account for what's actually different about these particular devices.

Step 2: Ask Copilot to diagnose the patternCopilot

Have it look for what the failing devices have in common — OS version, device group, network segment — rather than treating each failure as unrelated.

Prompt idea:

These 12 devices failed this app's deployment with error [code]. What do they have in common, and what's the most likely root cause based on that pattern?

Step 3: Validate the diagnosis against a couple of devices manuallyCopilot

Check the proposed root cause against 2-3 actual devices before rolling out a fix fleet-wide — a plausible-sounding pattern can still be coincidental.

Step 4: Roll the fix out in stages, not all at onceCopilot

Deploy to a small batch first and confirm success before pushing to every affected device.

Check the work

• The diagnosed root cause was confirmed against actual devices, not just inferred from the pattern.
• A staged rollout confirmed the fix before full deployment.
• No previously-working device was broken by the fix.

Source: Microsoft Copilot Scenario Library — Manage application installations (2026)