Show HN: Estimating what users ask ChatGPT about your company
An open-source pipeline reverse-engineers likely user prompts about a brand to audit why and how companies appear in ChatGPT and other LLM responses. It crawls websites, analyzes search keywords, and uses WildChat data to estimate what users ask, filling the analytics gap left by AI platforms that don't expose query data.
Background
Companies increasingly rely on AI chatbots (ChatGPT, Gemini, Perplexity) as a discovery channel, but unlike traditional web search — where Google Search Console tells you what queries drive traffic — there is no dashboard showing which prompts lead a chatbot to mention your brand. This project attempts to reverse-engineer that black box by crawling your site for keywords, mining WildChat (a public dataset of ~1M real ChatGPT conversations), and synthesising the most probable user prompts that would cause an AI to cite you or a competitor. The result is essentially a proto-analytics tool for a world where brand visibility shifts from search-engine rankings to AI-generated answers.