Quick Reference

Platform comparison and prompting cheat sheet

Platform Comparison

All three platforms offer free tiers, but the paid plans ($20/month each) are where the real capability lives — current flagship models, longer context windows, and features like deep research and file creation. If you’re serious about seeing what AI can actually do, the paid tier is the starting line, not the upgrade.

Feature	Claude	ChatGPT	Gemini
Free tier	Yes	Yes	Yes
Paid plan	$20/month (Pro)	$20/month (Plus)	$20/month (Advanced)
Web search	Built-in	Built-in	Built-in (Google)
Deep research	Yes	Yes	Yes (Advanced)
File uploads	Yes	Yes	Yes
Image generation	No	Yes (DALL-E)	Yes (Imagen)
Voice conversations	Limited	Yes	Yes
File creation	Word, PDF, spreadsheets, slides	Via Canvas + code	Google Docs export
Workspaces	Projects	Projects	Gems
Memory	Automatic + manual	Automatic + manual	Saved Info + auto
Private mode	Incognito	Temporary Chat	Pause Activity

Privacy Settings — Quick Setup

Platform	What to do	Where
Claude	Review privacy settings	Settings → Privacy
ChatGPT	Toggle off “Improve the model for everyone”	Settings → Data Controls
Gemini	Pause Gemini Apps Activity	myactivity.google.com

Personalization — Quick Setup

Platform	What to do	Where
Claude	Set user preferences	Settings → Profile
ChatGPT	Set Custom Instructions	Settings → Personalization
Gemini	Add Saved Info	gemini.google.com/saved-info

Prompting Cheat Sheet

Start Strong

Bad: “Help me with a thing.”

Good: “I need to [specific task]. Here’s the context: [details]. The audience is [who]. I want the tone to be [how]. Keep it under [length].”

The more specific your request, the better the output. Details aren’t optional — they’re the difference between generic and useful.

Useful Follow-up Patterns

What you want	What to say
Shorter	“Cut this in half. Keep only the essential points.”
More specific	“The section on [X] is too vague. Add concrete examples.”
Different tone	“Rewrite this for [audience]. It should sound [tone].”
Reorganized	“Move [section] before [section]. The current order doesn’t flow.”
Simplified	“A smart 12-year-old should understand this. Rewrite accordingly.”
More critical	“What’s wrong with this? What am I missing? Be honest.”
Different angle	“Now argue the opposite position with the same rigor.”
Self-check	“Review what you just wrote for errors, gaps, and weak reasoning.”

Role-Setting Phrases

Telling the AI to respond from a specific perspective changes what it prioritizes:

“You are an experienced [role]…”
“Respond as if you’re a [role] advising a client…”
“Think about this from the perspective of a [role]…”

When the Output Isn’t Good

Don’t accept mediocre results. Push back:

“That’s too generic. Be more specific to my situation.”
“You’re over-explaining. I already know [X]. Skip to the part I don’t know.”
“This doesn’t sound like me. Here’s how I’d actually say it: [example].”
“That fact doesn’t sound right. Can you verify it?”
“Start over with a different approach. That angle isn’t working.”

What AI Is Good At (and Not)

Strong

Drafting and editing text (emails, reports, documents)
Explaining and teaching concepts at any level
Brainstorming ideas and options
Organizing messy information into structure
Analyzing tradeoffs and decisions
Summarizing long documents
Translating between audiences (technical to non-technical, etc.)
Writing and explaining code

Weaker

Precise math beyond basic arithmetic (it can get calculations wrong)
Very recent events (web search helps, but verify)
Highly specialized or niche professional knowledge (it’s broad, not always deep)
Predictions and forecasts (it doesn’t know the future)
Anything requiring access to your personal systems or accounts

Needs Caution

Factual claims (verify anything you’ll act on)
Legal, medical, or financial advice (use as a starting point, not a final answer)
Sensitive personal information (don’t share what you wouldn’t want stored)