Claude isn't 'just a coding AI' — 13 prompt patterns for research, learning, and decisions

Series part 7

Ever found yourself stuck on a decision and realized — Claude could help, but somehow you’ve only been using it for code?

Part 6 covered Claude Code’s unique features. This post shifts gears: using Claude for tasks that don’t involve writing code.

Claude tends to get described as “the coding AI,” but it’s equally strong at research, learning, and decision-making. I rely on it daily as a “thinking partner” — and honestly, that’s the use case that surprised me most.

B. Research & learning

1 — Technology survey

Please research {{technology}} from these angles:

- Overview (3 lines)
- Comparison with at least 3 similar technologies
- Pros / cons of adoption
- Well-known adopters and use cases

2 — Reading official docs

Please read this official documentation and tell me:
{{URL or pasted content}}

- The 5 core concepts
- Points where beginners typically get stuck
- A 30-minute starter project idea

3 — Drafting a study plan

I want to achieve {{goal}} in {{timeframe}}. Please draft a study plan.

- My current level: {{experience and knowledge}}
- Weekly milestones
- Resources and deliverables per week

4 — Comparison

Compare {{A}} vs {{B}} vs {{C}} across:
- Learning curve / operational cost / ecosystem / real-world adoption

Output: matrix + the most balanced option on each axis with reasoning

5 — Checking your understanding

Here's my current understanding of {{concept}}:
{{bullet list of how you understand it}}

Please point out:
- What I have right
- What I have wrong
- Important pieces I'm missing

This is the killer pattern. Voicing your current mental model and having Claude audit it exposes misconceptions fast.

6 — Case studies

Find 5 real-world examples of solving {{problem}}.

For each:
- Company (within public info)
- Stack used
- Key decision drivers
- Measurable outcomes

7 — Steelman the opposition

List at least 5 counterarguments to {{my claim}}.

- Reasoning behind each
- Which counterargument is strongest
- What additional evidence would strengthen my claim

Use weekly to find blind spots.

C. Thinking & decisions

8 — Issue triage

Read this discussion and extract the issues:
{{minutes / transcript}}

Output:
- Up to 5 issues
- Arguments for / against each
- Unresolved items and the decision-maker for each

9 — Decision framework

Evaluate the following decision:
{{decision}}

- Pros (5+)
- Cons (5+)
- Cost of doing nothing
- Is it reversible or irreversible?
- How will 1-year-future me see this?

“How will 1-year-future me see this?” snaps you out of short-term thinking.

10 — Debate

I hold position {{A}}. You take position {{B}} and argue against me.
Do 3 rounds. At the end, summarize which side was stronger on which axis.

11 — Summarization

Summarize the long text below:
{{long text}}

Three levels:
- One-line summary (≤140 chars)
- Three-line summary
- 20-line summary (cover all key points)

12 — Applying frameworks

Analyze {{problem}} using these frames:
- MECE decomposition
- 5W1H
- Before / After

13 — Metacognition

I've been stuck on {{worry}}.
- What problem am I actually trying to solve?
- What is the essence of this worry?
- What decision am I postponing?

When you’re looping in your own head, Claude is willing to tell you “the premise of your question might be wrong” without flinching.

Tips

Tip	Why
Always specify output format	Thinking-mode prompts get vague answers without it
Always include counter-arguments	Claude tends to flatter; force it to push back
Chain patterns	Pattern 5 (understanding) → 7 (counterarguments) → 9 (decision) is a killer combo

My take — Claude as a “thinking warm-up”

I used to struggle to reach conclusions when discussing tech decisions with teammates. Now I debate with Claude first.

Claude counterargues my position → blind spots emerge
Claude compares candidates → evaluation axes become objective
Check decisions with a “1-year-future” lens → short-term bias melts

The upshot: my conversations with teammates start from a much sharper set of questions. Claude isn’t a replacement; it’s a warm-up partner for thinking.

If you’ve got one decision sitting heavy in your head right now, you might try Pattern 9 (the “1-year-future me” lens) on it tonight. Often, the answer was already obvious once the timeframe stretched.

Claude Code-only features — tooling edition
Treat Claude like a brilliant new intern — principles
Reference-grade Qiita version: Qiita profile