Build a real FAQ with Perplexity instead of guessing what users ask
Pull actual questions from the web and your own docs, then let it draft answers you verify once.
Most FAQs are written by the people who built the product, so they answer questions nobody asked. Perplexity helps you start from what people actually search.
I ask: 'What are the most common questions people ask about [topic] according to forums and help centers?' It returns sourced questions, and the sources tell me which ones are worth answering publicly.
Then I feed it my own docs and say 'answer each question using only this material.' That keeps the reply grounded in what we actually ship instead of the internet's guess.
Every answer gets one human read before publish. A sourced FAQ drafted by an AI but verified by a person beats a handwritten one full of assumptions.