# Prompt ka formula: Goal, Input, Format

PROMPT · Last check: 2026-11-17

'Notes bana do' ki jagah aisa prompt likho jo exam-ready notes de, kisi bhi AI chat mein.

- **Kya paid hai:** Kisi bhi free AI chat mein try kar sakte ho (Claude Free, ChatGPT Free, Gemini). Free plan ki usage limit hoti hai.
- **Phone pe?:** Haan. Kisi bhi AI chat app ya browser mein chalega.
- **Kahan chalega:** Koi bhi device. Humne Claude Sonnet 5.5 pe test kiya.
- **Time:** 5 minute

## Kyun kaam ka hai
Jab prompt mein goal aur format nahi hota, AI guess karta hai. Goal, asli content aur output ka format clear do, toh jo chahiye wahi milta hai, aur revision aasaan ho jata hai.

## Kya chahiye
- Koi bhi AI chat account (free chalega).
- Tumhara asli content: textbook paragraph ya class notes.

## Steps
1. **Goal likho**: Ek line: kis kaam ke liye? Example: `Goal: kal ke exam ke liye 2 minute revision notes.`
2. **Input do**: Paragraph poora paste karo. AI wahi use karega jo tum doge. `Input (paragraph neeche): ...`
3. **Format batao**: Output kaisa chahiye, exactly. Example: `Output format: 1 line definition, phir exactly 4 bullet points (har point 12 words se kam), aur end mein 1 likely exam question.`
4. **Language bolo**: Example: `Simple Hinglish, technical words English mein.` Isse terms wahi rehte hain jo exam mein aate hain.
5. **Book se match karo**: AI galti kar sakta hai. Har point ko apni book ya notes se ek baar check karo.

### Revision notes template
```
Goal: [kis kaam ke liye, jaise kal ke exam ki revision]
Input (neeche):
[PARAGRAPH PASTE KARO]
Output format: [jaise 1 definition + 4 bullets + 1 exam question]
Language: simple Hinglish, technical words English mein.
```

### Humara exact test prompt (2)
```
Goal: kal ke exam ke liye 2 minute revision notes.
Input (paragraph neeche):
CPU cache ek chhoti, bahut fast memory hoti hai jo processor ke paas hoti hai. Jab CPU ko koi data baar-baar chahiye hota hai, to woh RAM tak jaane ke bajaye cache se le leta hai, isliye kaam fast hota hai. Cache ke levels hote hain: L1 sabse chhota aur sabse fast, L2 thoda bada, aur L3 sabse bada lekin thoda slow. Agar data cache mein na mile, to use cache miss kehte hain.
Output format: 1 line definition, phir exactly 4 bullet points (har point 12 words se kam), aur end mein 1 likely exam question. Simple Hinglish, technical words English mein.
```

## Result kaisa aana chahiye
Ek definition, exactly utne bullets jitne maange, aur ek exam question, sab tumhare paragraph se.

## Agar kaam na kare
- **Bahut lamba answer**: Format mein number daalo: "exactly 4 bullets, har point 12 words se kam".
- **Book se alag fact**: Likho: "Sirf mere paragraph ki baatein use karo, kuch extra mat jodo."
- **Pura English ya pura Hindi aa gaya**: Language line mein example do: "jaise: 'Cache fast memory hai jo CPU ke paas hoti hai'".

## Free vs paid: asli catch
- Har run ka output thoda alag ho sakta hai; humare 2 runs milte-julte the, same nahi.
- Free plan mein ek time mein messages ki limit hoti hai.
- Humne ek topic aur ek model pe test kiya. Ye demo hai, benchmark nahi.

**Safety:** Personal details, passwords ya ID numbers AI chat mein paste mat karo.

## Doosre options
- AI se pehle sawaal puchwao: "Notes banane se pehle mujhse 3 sawaal pucho."
- Apne purane achhe notes paste karke bolo: "Isi style mein banao."

## Sources (last check date pe dekhe)
- [Claude plans (free plan aur limits)](https://claude.com/pricing)
- [Anthropic prompt engineering overview](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/overview)

---
Riya ek virtual host hai. Workflows humari human team research aur test karti hai; har guide pe last check ki date hai.
https://riya-explains.pages.dev/g/prompt-ka-formula/
