AI Prompt Generator

What is the AI Prompt Generator?

The AI Prompt Generator is a free tool that writes 10 prompts a buyer might type into an AI assistant such as ChatGPT, Claude or Perplexity while researching your category. Each is a full sentence with real constraints like budget or team size, labelled with its angle, and ready to run as a check on whether AI answers name you.

How to use the AI Prompt Generator

1. Enter what you sell in Category or topic, described the way a buyer would, such as "AI visibility tracking".
2. Press Generate prompts.
3. Read the 10 Generated Prompts and the angle under each, such as comparison, recommendation or troubleshooting. The first time, sign up free to reveal them.
4. Copy one prompt, or press Copy All to paste the set into a sheet or tracker.

How is an AI prompt different from a search keyword?

A search is two or three words with the context stripped out. A prompt is a sentence with the context left in: the budget, the team size, the thing that already failed. Compare "best AI visibility tool" with "we are a four-person marketing team already paying for two SEO tools, what should we use to see whether ChatGPT recommends us".

The engine answers the whole sentence. The constraints inside it decide which products get named, and a product that fits the constraint can be recommended over a better-known one that does not. For the short, keyword-shaped strings instead, use the AI search query generator.

What should you do with the generated prompts?

Run each one in the engines your buyers use and record three things: whether you were named, where in the answer, and which sources the answer leaned on. The third column is the one you can act on. The pages an engine cites are the pages worth influencing, and they are usually roundups and comparison posts rather than your own site.

Do it on a schedule rather than once. One run is a sample, not a reading. For a quick automated sample, the AI Brand Visibility Checker writes five buying questions for your category and asks them for you.

Why do AI answers change between runs?

The same prompt returns different answers on different runs, to different accounts, on different days. Sampling explains some of it, retrieval more (the pages found today are not the pages found last week) and personalisation the rest.

So track a rate: out of ten runs of this prompt, how many named you. A single mention is not visibility, and a single absence is not a problem.

What do AI prompt examples for buyer research look like?

Buyer research prompts are full questions with the buyer's situation built in. For a category such as project management software, the generator's angles produce prompts like these. Comparison: "We are a 12-person design agency moving off spreadsheets. How does Asana compare with Trello for client work?" Recommendation: "What project management tool would you suggest for a remote team of 30 that needs time tracking?" Troubleshooting: "Our team stopped updating tasks in our project tool after two months. What tools make that less likely?" Worth it: "Is a paid project management tool worth it for a team of five?"

Each of those can be answered with a short list of products, which is exactly what you want to measure. The 10 prompts you get are labelled with their angle, so you can check that your set covers more than one kind of question.

Good buyer prompts leave the engine something to decide. "Is Asana good?" invites a description; "which tool should we pick" invites a recommendation, and the recommendation is what tells you whether you are in the running.

If your category has distinct buyer types, such as freelancers and enterprise teams, run the generator once per type by naming the type in the category box, for example "AI visibility tracking for agencies". The constraints written into the prompts should then suit that buyer.

Can this AI prompt generator write image or video prompts?

No. This AI prompt generator writes the questions buyers type into chat assistants, to test whether those assistants recommend you. It does not write prompts for image or video models, and it does not write task prompts that tell a model what to produce.

For a task prompt, such as an email, a brief or a code fix, use the best AI prompt generator, which builds one on a framework, or the generator for the engine you use. For images and video, a tool built for those models will serve you better, because they want a description of the scene rather than a question.

Searches for an AI image prompt generator usually mean a tool that turns an idea or a reference picture into a detailed description for an image model. That is a different job with different rules: describe the subject, style, composition and lighting, and say what to leave out. Video models want a shot description in the same spirit, with the action, camera movement and duration added.

How do you turn the prompts into a tracking sheet?

Press Copy All and paste the 10 prompts into a sheet, one per row, keeping the angle beside each. Add columns for the engine, the date, whether you were named, your position in the answer, the competitors named and the sources cited.

Run each prompt in a fresh chat so earlier conversation does not colour the answer, and run the important ones more than once. Record what came back as written, not your summary of it; the exact wording is what you will compare against next month.

After two or three rounds the sheet shows which angles you win and which you lose. Losing comparison prompts usually points to missing comparison content. Losing troubleshooting prompts usually points to help content the engines cannot find.

Keep the prompt set fixed for at least a quarter. Adding and removing prompts every month makes your rate move for reasons that have nothing to do with the engines. When you do change the set, note the date in the sheet.

A simple score works: for each prompt, the share of runs that named you. Average those by angle and compare the averages month to month.

Which AI engines should you test the prompts in?

Test in the engines your buyers actually use, which for most B2B categories means ChatGPT, Google's AI answers, Perplexity and Gemini, with Claude and Grok where your audience leans that way. The engines differ in how they answer: Perplexity searches the web for every answer and cites what it read, while chat assistants answer from what the model learned in training unless they decide to search.

That difference changes what you can influence. Where an engine searches, the pages it cites are the lever. Where it answers from training, you are working on a slower signal. The ChatGPT visibility checker runs a check in ChatGPT for you, and the other engine checkers do the same for theirs.

Use a fresh or signed-out session where you can. Saved memories, custom instructions and earlier chats can all shape an assistant's answer, so an account with your own company in its history is a poor stand-in for a buyer. Record the date and, where the interface shows it, whether the engine searched: the same prompt can come back with sources one day and without them the next.