The best prompts for Perplexity

Perplexity is an answer engine from Perplexity AI. It differs from a general chat assistant in one important respect: it is built around searching the web and writing an answer with links to the pages it drew on. That design changes what a good prompt looks like.

With a chat assistant, you typically supply the material and ask for it to be transformed. With Perplexity, you are usually asking it to find the material. The wording of your question therefore does double duty. It tells the product what to write and it shapes what gets retrieved. A vague question should be expected to retrieve vague pages. A question that names the subject, the time frame, the geography and the kind of evidence you want gives the search step something to work with.

This is why the prompts on this site list Perplexity only for research-with-sources tasks, such as market research, literature scans and competitor analysis. For drafting or editing text you already have, the search step adds little.

The long structured prompts in this library still work here, but the parts that matter most are different. Role descriptions and tone instructions carry less weight than a tightly scoped question, explicit rules about acceptable sources, and an output format that keeps each claim next to its citation. Follow-up questions in the same thread are a normal part of the workflow: start broad enough to see what exists, then narrow.

Citations are a starting point for verification and should not be treated as proof. A linked page may not say exactly what the summary attributes to it, so open the sources behind anything you plan to rely on.

Product features, modes and plans change often and are not described in detail here. Consult Perplexity's own help and documentation for current specifics.