> ## Documentation Index > Fetch the complete documentation index at: https://docs.surnex.io/llms.txt > Use this file to discover all available pages before exploring further. # April 2026 > AI search visibility — AI Overviews, ChatGPT, LLM benchmarking, and GEO. ### Citation gap Compare your domain against up to ten competitors across up to fifty keywords, and see every query where they're cited in an AI answer and you aren't. Gaps are flagged in the grid, and rows where you do appear show what you were **co-cited with**. Co-citation is the underused half. If the same publications and comparison sites keep appearing alongside you, those are the sources shaping answers in your market — and being covered by them moves the needle further than editing your own page. Competitor domains can be pulled from the project's configured list in one click. ### Benchmark across platforms Run one query against several AI platforms at once and see which cite you, how many sources each drew on, and which model produced the answer. Expanding a row gives the full response and its cited sources. The pattern across platforms is the finding. Cited nowhere is a content problem; cited on some platforms only means they weight sources differently, and comparing which ones cite you gives you something specific to test. ### AI keyword trends Search volume, velocity, and year-on-year growth for keywords in AI search — a different quantity from traditional search volume, and the divergence is the point. A term with modest search volume but high AI volume is one people increasingly ask an assistant rather than type. ### ChatGPT visibility Query ChatGPT and get back the response plus a table of the sources it drew on — source page, domain, the **snippet** it extracted, and relevance scoring. The snippet column is the most useful thing here: when a competitor is cited, it shows exactly what they said that you didn't. ### Google AI Mode The same for Google's conversational AI Mode, with the sources it cited. AI responses are non-deterministic. The same query can return a different source mix on different runs, so a single absence is a sample rather than evidence. ### AI Overview tracking Rank checks now record whether Google returned an AI Overview for each tracked keyword, and whether your domain was cited in it. No configuration — it comes from the checks you're already running. Two rates come out of it: * **Trigger rate** — the share of your keywords producing an AI answer at all. Exposure, not performance * **Citation rate** — of those, how many cite you. The number you can actually move Both surface on the ranking overview, the project dashboard, and as a badge on individual keyword cards. The keyword table adds a **references** count and the referenced sources per row. The rows worth attention are the ones where you rank well and aren't cited — the answer is being given without you, and your position is delivering less than it looks. ### Generative engine optimization Topic-level tracking of your presence in AI answers, on a daily schedule, so you can watch a subject area over time rather than checking one query at a time. Add the topics you want to be the answer for, phrased as questions someone would actually ask. Each snapshot records mentions, citation rate, AI volume, the model that answered, and a quality score for the response. **Mentions and citations are different things.** Being mentioned without being cited means you have brand awareness but the model isn't treating your site as the source — a content problem rather than an awareness one. ### AI competitors Analyze a topic and see which domains dominate the AI answers for it. Frequently not your search competitors — publications, documentation, forums and comparison sites appear far more often than vendors, which changes where the work is. ### Topic detail Snapshot history, the sources cited for a topic, and a viewer for the raw LLM response.