Free tool

AI Prompt Builder

Fill in the blanks, and the builder assembles a structured prompt you can paste into any AI chatbot. Nothing you type leaves your browser.

Build your prompt

Try an example:

Helpful extras

Your prompt

How to use the prompt builder

Most disappointing chatbot answers come from prompts that leave too much to guesswork. If you type “write a product description”, the model has to invent the product, the audience, the length and the tone — and it will usually pick something bland and middle-of-the-road. The builder walks you through the handful of details that remove that guesswork. You don’t need to fill in every box; even adding a clear task and one line of context makes a visible difference.

  1. Start with the task. Begin with a verb: write, summarise, compare, explain, review, plan. Be specific about the deliverable (“a 5-point checklist”, “three subject lines”, “a reply under 120 words”).
  2. Add context. Paste the facts the model can’t know: your product details, your notes, the text to edit, the policy you need to follow. Context is where most of the quality comes from.
  3. Set a role and audience. “You are an experienced tax accountant” and “for a first-time freelancer” change vocabulary, depth and assumptions.
  4. Choose a format and tone. Tables, bullet points, word limits and tone words give you an answer you can use without heavy editing.
  5. Add rules. Constraints like “don’t mention pricing” or “use British spelling” prevent the most common follow-up corrections.
  6. Copy and paste the result into ChatGPT, Claude, Gemini, Copilot or any other chatbot. Then iterate: if the answer misses, change one field and try again.

Why this structure works

Every major AI provider’s prompting documentation gives similar advice: be clear and direct, provide context, show examples, and say what format you want. The builder simply turns that advice into a form. Separating the parts also makes it easier to see what is missing. If an answer is too generic, the context field is usually thin. If it’s the wrong length or shape, the format field needs work. If it uses the wrong level of detail, adjust the audience.

The optional extras are there for good reasons. “Ask me clarifying questions first” stops the model from charging ahead on a wrong assumption — useful for anything personal, like a cover letter or a meal plan. “Reason step by step” can help with multi-step problems such as calculations or planning. “Say if unsure; don’t invent facts” nudges the model to flag uncertainty instead of filling gaps with plausible-sounding guesses. None of these are magic, but together they make answers more useful and mistakes easier to spot.

Three worked examples

1. A customer-service reply

Without structure: “Reply to an angry customer about a late order.” With structure: a role (customer-service manager), the facts (courier delay, free shipping offer), the audience (a polite repeat customer), the format (email under 150 words with a subject line) and a rule (don’t blame the customer). The second version produces a reply you could send after a quick read, because the model is no longer guessing what you can offer.

2. Study help that doesn’t do your homework for you

Set the role as a patient tutor, describe exactly what you find confusing, and add the rule “don’t give quiz answers until I reply.” You get an explanation pitched at your level and active practice, which is far more useful for an exam than a finished answer.

3. A code review

Give the language, what the code is meant to do and the symptom (“crashes on empty files”). Ask for issues ranked by severity and one-line explanations. You’ll get focused feedback instead of a full rewrite you didn’t ask for.

Editor’s note — add your own experience: Add 1–2 screenshots of a real before/after prompt you tested in at least two chatbots, and describe in your own words what changed in the answers. This first-hand evidence is what makes the page stand out.
Remove this box by adding your content here, or hide all notes with "showEditorNotes": false in site.config.json.

Tips for better results

  • Put the main instruction first. In long prompts, the task near the top is less likely to get lost.
  • Separate instructions from material. Label pasted text clearly (for example “TEXT:” or “NOTES:”) so the model knows what to work on and what to obey.
  • Give an example of what “good” looks like. One short sample of the style you want often beats a paragraph of description.
  • Change one thing at a time. When iterating, adjust a single field so you learn what actually made the difference.
  • Keep private data out. Don’t paste passwords, ID numbers or other people’s personal information into any chatbot. Read our AI privacy guide for the details.
  • Save prompts that work. Reusable prompts are a real time-saver. Browse our prompt library for 60 ready-made starting points.

Limitations

A better prompt improves the odds of a good answer; it doesn’t guarantee one. Chatbots can still misunderstand, make mistakes or state wrong facts with confidence. For anything important — health, legal, financial or academic work — check the answer against reliable sources. Our guide on how to fact-check AI answers shows a quick routine. The token estimate shown under the prompt is a rough rule of thumb; use our token counter for a fuller breakdown.

Frequently asked questions

Does this prompt builder work with ChatGPT, Claude and Gemini?

Yes. The output is plain text, so you can paste it into any chatbot. The structure — role, task, context, format and rules — is a general technique, not a trick for one specific model.

Is my prompt saved or sent anywhere?

No. The builder runs entirely in your browser. Your last draft is stored only in your own browser’s local storage so you don’t lose it on refresh. Click “Start Over” to delete it.

Do I need to fill in every field?

No. Only the task is really needed. Each extra field adds clarity, so fill in the ones that matter for your request and leave the rest empty.

Why does the builder add “say if you are not sure”?

Chatbots can produce confident-sounding mistakes. Asking the model to flag uncertainty and not invent sources doesn’t eliminate errors, but it makes it more likely you’ll notice them. Always verify important facts.

How long should a prompt be?

As long as it needs to be to remove ambiguity. Many good prompts are 50–200 words. Very long prompts are fine for complex tasks, but put the most important instruction near the start.

Sources

  1. Anthropic — Prompt engineering overview (opens in a new tab)
  2. OpenAI — Prompt engineering guide (opens in a new tab)
  3. Google — Prompt design strategies for Gemini (opens in a new tab)

Facts, prices and features were checked against these sources on November 19, 2026. AI products change quickly; if you spot something out of date, please tell us.