There's a gap opening between people who use AI tools and people who use them well.
The first group asks questions and gets answers. The second group uses AI to compress hours of work into minutes — research, writing, analysis, planning, coding — while keeping their own judgment at the center.
Here's how to join the second group.
Shift Your Mental Model First
Most people treat AI like a smarter Google — a tool you query for a single answer. This limits its usefulness dramatically.
The better frame: AI is a tireless first-draft collaborator. It doesn't replace your expertise or judgment, but it eliminates the blank-page problem and compresses the low-value work in almost every task.
Once you adopt this frame, the applications become obvious.
The Most Effective Use Cases
1. Writing (Any Kind)
Don't ask AI to write for you. Ask it to write with you.
Give it context, your angle, your key points — and let it produce a structured first draft. Then edit, refine, and add your voice. The result is often 70–80% of a final piece produced in 10% of the time.
High-value applications:
- Email drafts (especially difficult ones)
- Performance reviews and feedback
- Reports, proposals, and summaries
- Job applications and cover letters
Prompt that works: "I need to write [document type] for [audience]. My main points are [X, Y, Z]. Write a first draft that's [tone: direct/professional/friendly]."
2. Research and Synthesis
Instead of reading 10 articles to form a view, paste them into a long-context AI and ask it to synthesize, compare perspectives, and highlight the most important points.
Claude and Gemini both support very long context windows — you can paste entire reports and ask targeted questions.
Prompt that works: "Here are [3 articles / a report / a dataset]. Summarize the key findings, identify where they agree and disagree, and flag anything I should be skeptical of."
3. Thinking Through Decisions
When facing a complex decision, explain the situation to an AI and ask it to steelman the opposing view, list second-order consequences, or play devil's advocate.
This is one of AI's most underused capabilities. It's a pressure-free sounding board that doesn't have skin in the game.
Prompt that works: "I'm considering [decision]. Here's my current thinking: [X]. What are the strongest arguments against this? What am I likely not seeing?"
4. Learning Faster
AI tutors are genuinely superior for many types of learning. They adapt to your level, answer follow-up questions instantly, and never get impatient.
When you encounter something you don't understand — a financial concept, a piece of code, a technical term — ask AI to explain it as if you're a smart person who doesn't know the field. Then ask it to give you three examples. Then ask what the most common misconceptions are.
This three-step pattern builds faster understanding than reading alone.
5. Automating Repetitive Work
If you do the same type of task more than twice a week, there's probably an AI workflow that reduces it by 80%.
Examples:
- Summarizing meeting notes into action items
- Converting raw data into formatted tables
- Translating lengthy emails into one-line summaries
- Generating social media posts from longer content
The Prompting Principles That Make the Difference
Be specific about format: "Give me a bulleted list" vs. "Explain this" produces very different results.
Provide context: AI performs significantly better when it knows who you are, what you're trying to do, and who the output is for.
Iterate, don't regenerate: If a response is 70% right, tell it what to fix rather than starting over. Each iteration sharpens the output.
Ask it to check its own work: For factual claims, end with "What should I verify before relying on this?" This triggers a more careful self-audit.
Which Tools to Use
| Tool | Best for |
|---|---|
| Claude (Anthropic) | Long documents, nuanced writing, analysis |
| ChatGPT (OpenAI) | General use, image generation (GPT-4o) |
| Gemini (Google) | Research with web access, Google Workspace integration |
| Perplexity | Research with citations |
| Cursor / GitHub Copilot | Code |
You don't need all of them. Pick one primary tool and learn it deeply before adding others.
The Productivity Ceiling
The people getting the most out of AI are not the ones using it most. They're the ones who know which tasks to hand off and which to keep.
Use AI to eliminate low-value work — formatting, first drafts, research summaries. Keep the judgment, relationships, and creative decisions human. That combination is what compounds.
