The LLM SEO tool that turns AI answers into a plan you can ship

LLM SEO is the work of getting large language models to mention and cite you when they answer questions in your category. It is also called generative engine optimization (GEO) or answer engine optimization.

The goal shifts from a position on a results page to a place inside an answer. A buyer who asks ChatGPT for the best tool for their problem reads one paragraph and a few links. You are either in that paragraph or you are not. Our guide to generative engine optimization covers the theory; this page covers the tool.

A complete LLM SEO tool covers four jobs: check that AI engines can read your site, measure what they say, decide what to change, and show whether the change worked. Tools that only track mentions leave the middle two to you.

It reads your site the way AI crawlers do and reads the answers the way your buyers do, then compares the two. Most gaps come from one of three causes: the crawler is blocked, the page has nothing quotable, or third-party pages the engine trusts list your competitors and not you.

On your site

The GEO audit fetches your key pages without running JavaScript, through your robots.txt and response headers. It lists 14 AI crawlers and what your robots.txt tells each one, and it also requests your home page with their user agents to catch a CDN or firewall that blocks them. An AI read scores each content page from 0 to 10 for answerability and entity clarity, and flags pages with no direct answer or nothing to quote.

In the answers

AI visibility asks your prompts in the real products, ChatGPT with web search on, Gemini and Google AI Overviews, from your market's country. It keeps every source, then opens the most-cited third-party pages to see who they mention. A Get listed learning names a cited page that lists competitors and not you.

LogNorm generates the files and pages that fix the most common GEO gaps, built from your crawled pages, your Company Brain and competitor data. You copy them into your site, or your coding agent ships them for you.

• llms.txt and llms-full.txt, a markdown map of your key pages for AI assistants.
• A robots.txt AI section with explicit, commented rules for each AI crawler.
• An agent discovery kit: API catalog, Link headers, Content Signals and markdown-for-agents setup.
• A JSON-LD bundle with Organization, WebSite, SoftwareApplication and FAQPage snippets.
• Answer-first pages: a category definition page, an FAQ page, a comparison page, a glossary and a fact sheet for AI assistants.
• Quotable rewrites for the pages AI engines find hardest to cite.

Page drafts can contain TODOs where LogNorm lacked a fact. Fill them in before publishing, because an invented fact is the fastest way to lose an engine's trust.

Write prompts the way a buyer asks before they know your name: best invoicing software for a small agency, not is Acme good. Branded prompts only tell you what you already know.

Suggest prompts in LogNorm writes them from your positioning, topics and tracked competitors, and labels each with its kind (category, alternatives, how-to, comparison or evaluation) and buying stage. Prompts you add yourself are filed under the closest topic, so AI visibility work lines up with your keyword and content plan.

Before the first run, set your brand identity: product names, former names and abbreviations count as mentions, and owned sites such as your docs domain or github.com/your-org count as citations. Saving it re-checks stored answers at no credit cost.

Yes. A Claude Code or Codex agent connected to LogNorm can review the answers, retire prompts that don't match real buyer questions, turn learnings into moves and ship GEO fixes such as llms.txt or schema in your codebase.

It works on the same plan as your team and leaves a trail of comments and decisions, and LogNorm re-checks its fixes on the live site. See the AI marketing agent page for how agents join the team.

Mostly in name. People search for a ChatGPT SEO tool when ChatGPT is where their buyers ask, but the same buyers also read Gemini and Google's AI Overviews, and each engine picks sources differently.

LogNorm keeps the engines apart. An Engine gaps learning appears when one engine ignores you while the others don't, so you can tell a ChatGPT problem from a site-wide one. If you only need per-engine tracking and share of voice, see the LLM rank tracker page.

Measure mentions and citations separately, per engine, over several runs. They answer different questions: a mention means the engine names you, a citation means it uses your page as evidence.

LogNorm links every published page and completed move to the prompts it was meant to affect and compares runs before and after the ship date. Verdicts read Improved, Cited, No change, Declined or Too early, and comparisons built on few answers are labelled a small sample. AI answers vary from one ask to the next, so one improved page is evidence, and the same pattern across several is something to plan on.