How to write AI prompts that get accurate answers

Follow these practical techniques to write AI prompts with clear context, defined scope, and output constraints that reduce hallucination and improve results.

The quality of an AI response depends almost entirely on the quality of the prompt you write. A clear, well-structured prompt gives the model the context it needs to stay on topic, answer accurately, and format its output usefully. These techniques will help you write prompts that leave less room for the model to guess — and less room for it to get things wrong.

Techniques for better prompts

1. Specify your context

Tell the model who you are or what this is for. Without this, the model assumes a generic audience and tailors its answer accordingly — which may not match your situation at all.

Vague	Better
”Explain contract law."	"I am a small business owner with no legal background. Explain the key clauses I should look for in a vendor contract.”

2. Define the scope

Tell the model how much ground to cover. Open-ended prompts produce unpredictably sized responses. Scoped prompts give you what you actually need.

Vague	Better
”Tell me about climate change."	"Give me a 3-point summary of the main causes of climate change, written for a general adult audience.”

3. State your constraints

Explicitly rule out what you don’t want. If the model doesn’t know what to exclude, it won’t exclude it.

Vague	Better
”Give me marketing copy for this product."	"Write marketing copy for this product. Do not use technical jargon. Do not make claims about pricing or availability.”

4. Describe the desired output format

Specify how the response should be structured. A model that doesn’t know your preferred format will pick one for you — and its choice may not fit how you plan to use the output.

Vague	Better
”Summarize this meeting transcript."	"Summarize this meeting transcript as a bulleted list of action items. Each item should name the person responsible and the deadline if one was mentioned.”

5. Flag your own uncertainty

If you’re not sure about something relevant to your question, say so. This tells the model to be careful in that area rather than confidently filling in the gap.

Vague	Better
”What are the tax implications of selling my business?"	"What are the general tax implications of selling a small business in the US? I’m not certain whether this would be treated as an asset sale or a stock sale, so please flag where that distinction matters.”

Before and after

Here is how these techniques combine in practice:

Before:

“Write something about onboarding.”

After:

“I am a product manager at a B2B SaaS company. Write a 200-word welcome email for new users who have just signed up but haven’t yet completed setup. The tone should be friendly but professional. Do not mention pricing or upsells. End with a single clear call to action directing them to the setup guide.”

The revised prompt specifies who is writing it, what it’s for, who it’s addressed to, the length, the tone, what to exclude, and what to end with. The model has almost no decisions left to make on your behalf.

Run your prompt through S.I.M.P. first. The issues and changes panels will tell you exactly where your prompt is falling short.