A Reusable Prompt Template for Clearer AI Answers

A short prompt can work well when the task is simple. For work where the answer needs to fit a real audience, source or format, the model needs enough context to make fewer guesses.

Use this structure as a checklist. You do not need every field for every request; add the parts that affect the result.

The prompt pattern

Task: [What should the assistant do? Start with a clear action verb.]
Context: [Who is this for? What are you trying to achieve?]
Source material: [Paste text, facts, links, or examples. Say if this is the only source.]
Requirements: [Length, tone, must-include details, exclusions, and boundaries.]
Output: [Format, sections, columns, or file-ready structure.]
Quality check: [What should it verify before returning the answer?]
If something is missing: [Ask a question or mark the gap; do not invent it.]

For a simple task, reduce the pattern to one or two lines. For a repeatable task, keep the fields that make the output consistent.

A worked example

Suppose you need a launch email for a small digital product. A vague request such as “write a persuasive email” leaves the audience, actual offer, length and proof points unstated. Try:

Task: Draft one launch email for the product described below.
Context: The readers are independent marketers who downloaded a free checklist last week.
Source material: Product: a 20-prompt campaign planning pack. Price: $17. Includes a PDF and editable worksheet. Do not add benefits that are not listed here.
Requirements: Under 160 words. Friendly and direct. No fake urgency, income claims, or phrases such as “game changer.”
Output: Subject line, preview text, email body, and one call to action.
Quality check: Confirm every feature and price matches the source material. If a fact is not provided, leave it out.

The prompt gives the model a real source of truth, a target reader and boundaries. The quality check is not a substitute for your review; it points attention to the details most likely to be fabricated or changed.

Improve the prompt in a second pass

Read the first answer and identify one specific failure. Was it too formal? Did it omit an important detail? Did it add a claim? Revise only the instruction that addresses that failure, then run it again.

For example, “make it better” gives little direction. “The first paragraph repeats the subject line. Replace it with a concrete description of the worksheet, and keep the price unchanged” gives the model a bounded edit.

If the assistant keeps missing the same thing, add a short positive example or show the structure you want. Make examples representative; an example that conflicts with the written requirements can confuse the result. Anthropic’s prompting guidance also recommends clear instructions, useful examples, and structure for complex prompts.

When to use this pattern

Use it for tasks you will repeat, work that has a specific audience, drafts tied to supplied facts, and outputs that need a predictable shape. Skip unnecessary role instructions when they do not help the task. “You are an expert” is less useful than telling the model what to do, what information it can rely on, and what a good result looks like.

Before using the output, check names, numbers, links, policy statements, quotes and other claims against their original sources. For a related method, see how to verify AI research before you use it.