The AI search playbook

when someone asks ChatGPT or Perplexity “what app should i use for this”, the assistant recommends from two things: what it can read on your own site and what the rest of the internet says about you. this playbook covers both levers, answer-shaped pages you control and third-party presence you earn, and it is honest about the fact that this channel pays slowly.

The shift that already happened

a growing share of product discovery now ends inside an answer instead of a results page. the user asks the assistant which app to use, the assistant names two or three products with a reason for each, and the conversation moves on. there is no page two and no ad slot to buy your way into most of those answers. the assistant recommends from what it can read: your pages, if they are crawlable and actually answer the question, and everything else written about you, Reddit threads, other people’s listicles, directory descriptions, reviews.

that second part matters more than most founders expect. an assistant treats your own claim about your own product as the weakest evidence available. what other people say about you, in text a crawler can retrieve, is what gets you named.

The two levers

there are only two levers, and you need both. lever one is your own site: pages built as direct answers to the questions your buyers actually ask, served as fast static HTML that a crawler can read without executing anything, with the answer visible near the top instead of buried under warmup prose. lever two is third-party presence: honest activity in the Reddit threads where your users already are, placement in other people’s comparison pages and listicles, and a small set of real directories that describe your product accurately.

lever one is fully in your control and moves first. lever two is slower and out of your hands, which is exactly why it counts as evidence. the Reddit side of lever two has its own complete method in the SAFE reddit playbook, because doing it wrong costs accounts, not just time.

Why start before it is obviously worth it

because everything in this channel compounds slowly. citations lag crawls, retrieval indexes lag publishing, and model training snapshots lag everything else by months. the product that started answering its market’s questions a year ago now sits inside the answers, and displacing it takes longer than joining it early would have. that is not urgency talk, it is just how lag works. the work is unglamorous, the feedback loop is measured in months, and the founders who win it are mostly the ones who started before there was proof it would pay.

This site is the experiment

i am not asking you to take any of this on faith, i am running it on the site you are reading. static HTML, AI crawlers explicitly allowed in robots.txt, every page written answer-first, and the whole setup published openly in how i run this site. it is early, and i do not have results to show you yet. i would rather tell you that plainly than dress up a chart, and as the numbers arrive, good or bad, they will be published the same way.

the rest of this playbook breaks the method into pieces: how assistants actually choose what to recommend, how to write pages that get quoted, where Reddit honestly fits, what technical work matters, and how to measure whether any of it is moving.

if you would rather have this mapped onto your app than read about it, the breakdown call is free. Let’s talk Growth.