How to write better AI prompts
Prompt writing is the skill that decides how useful AI is for you. This guide teaches a simple, repeatable framework plus practical techniques to get accurate, well-formatted answers from ChatGPT, Gemini and other AI tools.
The single biggest difference between people who find AI 'amazing' and those who find it 'useless' is not the tool — it is the prompt. A good prompt turns a vague, generic reply into a precise, ready-to-use answer.
The good news: prompting is a learnable skill, not magic. Once you understand the few patterns that consistently work, you can apply them to study, office work, content writing and your own business — in any language the AI supports.
This guide gives you a reusable framework and the techniques that matter most, with examples you can adapt for Nepali contexts.
The RTCF framework: Role, Task, Context, Format
Most strong prompts contain four parts. You do not always need all four, but adding them dramatically improves results: Role (who the AI should act as), Task (exactly what you want done), Context (the details and constraints), and Format (how the answer should look).
Example: 'You are an experienced accountant in Nepal (Role). Explain the difference between VAT and TDS to a small shop owner (Task). Assume they have no accounting background and operate a retail store (Context). Answer in simple English, under 200 words, with a short example for each (Format).' This will beat a bare 'what is VAT and TDS' almost every time.
Be specific and give examples
AI cannot read your mind. The more precisely you describe what 'good' looks like, the closer you get. Replace vague words with measurable ones: not 'short' but 'about 100 words'; not 'professional' but 'formal and polite, suitable for a government office'.
Showing an example is even more powerful than describing one. If you want product descriptions in a certain style, paste one good example and say 'write 5 more like this'. Giving the AI a sample to imitate is called 'few-shot' prompting, and it works remarkably well.
Iterate — treat it as a conversation
Your first prompt rarely gives the perfect answer, and that is normal. Instead of starting over, refine: tell the AI exactly what to change. 'Make it shorter.' 'Use simpler words.' 'Add a section on risks.' 'Rewrite the third point — it is wrong because…'
Because the chatbot remembers the current conversation, each correction builds on the last. Think of it like guiding a junior assistant: small, clear instructions get you steadily to the result you want.
Powerful techniques worth knowing
A handful of techniques reliably improve output quality:
- Ask it to think step by step for maths, logic or planning — it reduces careless mistakes.
- Ask it to ask you questions first ('Before answering, ask me anything you need to know') to fill gaps in your prompt.
- Request multiple options ('give me 3 different versions') and pick or combine the best.
- Tell it to adopt a perspective ('review this as a strict examiner' or 'as a cautious investor') for sharper feedback.
- Ask it to cite or flag uncertainty ('mark anything you are not sure about') so you know what to verify.
Common mistakes that ruin results
A few habits quietly sabotage your answers. Avoid these:
- Being too vague — one-line prompts with no context get generic replies.
- Asking for too many unrelated things in one prompt — split big jobs into steps.
- Trusting the answer blindly — always sanity-check facts, numbers and any local Nepali specifics.
- Forgetting to set the language — say 'reply in Nepali' or 'reply in English' if it matters.
- Not giving feedback — if the answer is off, say why, so the next one improves.
Key takeaways
- ✓Use the RTCF framework: Role, Task, Context, Format — adding these four elements transforms answer quality.
- ✓Be specific with measurable instructions, and show an example of what you want ('few-shot') whenever you can.
- ✓Treat prompting as a conversation: refine with small, clear corrections instead of starting over.
- ✓For reasoning tasks, ask the AI to work step by step and to flag anything it is unsure about.
- ✓Always set the output language and format, and verify facts before you rely on them.
How to Write Better AI Prompts — FAQ
What is a prompt in AI?+
A prompt is the question or instruction you give an AI tool. It can be a single sentence or a detailed brief with a role, task, context and desired format. The clearer and more specific your prompt, the more useful the AI's response.
Do longer prompts always give better answers?+
Not always — what matters is relevance, not length. A focused prompt with the right role, context and format beats a long, rambling one. Add detail that genuinely helps the AI understand the task, and leave out the rest.
Can I write prompts in Nepali?+
Yes. You can prompt in Nepali, English or a mix, and ask for the reply in whichever language you prefer. For complex tasks, results are often slightly more reliable in English, but everyday Nepali prompts work well.
What is 'few-shot' prompting?+
It means giving the AI one or more examples of the output you want before asking it to produce more. For instance, pasting a sample product description and saying 'write five more like this' usually gives much closer-to-target results than describing the style in words.
Sources & data note
These guides explain widely-accepted SEO, AEO and GEO practice as documented by Google Search Central, schema.org and current industry research. Search and AI systems evolve continually — treat specific thresholds (e.g. Core Web Vitals targets) as current guidance and verify against the latest official documentation. Examples are tailored to Nepal's market.