💡 AI tip: Start your prompt with what you want to accomplish, not a question.
Instead of: 'Can you help me with my resume?'
Try: 'Rewriting my resume. My goal is to to emphasize my leadership experience for a VP role to get this _____ specific job.'
💡 AI tip: Start your prompt with what you want to accomplish, not a question.
Instead of: 'Can you help me with my resume?'
Try: 'Rewriting my resume. My goal is to to emphasize my leadership experience for a VP role to get this _____ specific job.'
Your AI is only as good as the instructions you leave for it.
Most people treat AI like a search bar — ask, get answer, close tab. But you can give it a file of rules it reads before every session. What account to use. What tools are off-limits. What to check first.
Without it, the AI guesses. With it, the AI executes.
Think of it like onboarding a new employee: the better your documentation, the less they mess up on day one.
People, please, think of the environment!
Only print out your chatbot conversations if it is really necessary.
Did you know? Just highlight text. Copilot becomes a mind reader.
No file drag & drop required. Highlight to extract deep insights.
Stop attaching files manually for simple queries.
"Keep the structure, change the examples."
When AI nails the format but uses boring examples:
Don't rewrite from scratch. Tell it what to keep. Tell it what to change.
You're not starting over. You're refining.
Prompting is 20% of the value.
The other 80%:
→ Directing the output
→ Iterating on what you get
→ Organizing what works
→ Building reusable systems
Don't stop at the prompt. That's where it starts.
"What would a skeptic say about this?"
Ask AI to argue against its own output.
You'll find holes before someone else does. Makes your final work stronger.
Let AI be its own devil's advocate.
System prompts vs user prompts:
System prompt = standing instructions (always applies)
User prompt = single request (this conversation only)
Put your preferences in system. Put your task in user.
You don't have to repeat yourself every time.
"Show me 3 different versions of this."
One of the most useful follow-ups you can ask.
You're not stuck with version 1. You can always ask for options.
Pick the best. Combine elements. Or use one as a starting point.