Best Practices for Using AI for Tech Support Prompts

A simple explanation anyone at Tech2Success can follow

1. Start with a clear, direct problem statement

AI works best when you tell it exactly what’s happening.

Example:  “User cannot save files to SharePoint. Error: ‘You don’t have access.’ They are using Windows 11 and OneDrive.”

Why this matters: AI needs the symptoms, the environment, and the error message to give accurate guidance.

2. Include the user’s environment

Always specify:

• Windows version

• App (Outlook, OneDrive, SharePoint, Teams, etc.)

• Cloud tenant or domain

• Whether the issue is on the user’s PC or admin portal

This prevents the AI from giving instructions for the wrong platform.

3. Ask for step‑by‑step instructions

AI is strongest when you ask for ordered steps, not general advice.

Example:  “Give me step‑by‑step instructions to fix OneDrive sync errors for a user.”

This produces actionable guidance instead of vague suggestions.

4. Tell the AI what role you are in

AI responds differently depending on whether you’re:

• an admin

• a technician

• or helping an end user

Example:  “I am the admin. I need steps to verify SharePoint permissions.”

This ensures the AI gives admin‑level steps instead of user‑level steps.

5. Provide screenshots when possible

Screenshots help the AI identify:

• the exact screen

• the exact menu

• the exact error

• the exact context

This eliminates guesswork and speeds up troubleshooting.

6. Ask the AI to summarize or rewrite for customers

AI is excellent at turning technical explanations into customer‑friendly language.

Example:  “Rewrite this explanation so it’s easy for a non‑technical customer to understand.”

This saves time and improves clarity.

7. Use AI to verify your thinking, not replace it

AI is a support tool — not the decision maker.

Best practice:

• Use AI to generate steps

• You validate the steps

• You apply them safely in the customer’s environment

This prevents accidental misconfiguration.

8. Avoid giving AI sensitive data

Do not include:

• passwords

• private keys

• confidential customer information

• internal security tokens

AI doesn’t need these to troubleshoot.

9. Ask for SOP‑style outputs when documenting

AI can generate:

• KB articles

• SOPs

• checklists

• troubleshooting flows

• customer‑ready explanations

This helps standardize your documentation.