Your customers stopped searching. They started asking.

Generative Engine Optimisation is the practice of getting your brand named and cited inside AI-generated answers — in ChatGPT, Claude, Perplexity, Google AI Overviews, Gemini and Microsoft Copilot — rather than ranked in a list of blue links. It is also called LLM SEO, AEO (Answer Engine Optimisation) or LLMO. Cited measures your brand's mention rate across those platforms and fixes the content, structured data and entity signals that determine whether the AI says your name.

They share about 70% of the work — good content, technical hygiene, authoritative links. The other 30% is different: AI answers are synthesised from 3–7 sources, so there is no position #1, only included or excluded. LLMs reward pages that answer the question directly in the first 100–200 words, in plain prose, with named entities and verifiable claims. GEO also has its own measurement problem, which is why Cited exists: Google Search Console can't tell you what ChatGPT says about you.

In four stages. First, query fan-out: the AI decomposes the user's question into several independent sub-queries. Second, retrieval: it searches the web for each one — if you don't rank, you can't be cited. Third, extraction: it reads the retrieved pages and keeps passages that answer directly, in self-contained prose, with verifiable specifics. Fourth, synthesis: it writes the answer and names the sources it trusted. Cited gives you tools for every stage, including a fan-out simulator that shows the sub-queries your content must survive.

The headline metric is brand mention rate: for a curated set of real customer prompts, the percentage of AI answers that name your brand. Around it we track citation rate (is your domain actually linked), sentiment of each mention, and share of voice against named competitors — plus a hallucination watch that flags when an AI states something false about your brand, because a wrong answer at scale is worse than no answer.

For a small brand in a defined niche, being mentioned in 20–30% of relevant AI answers within 12 months is a strong, achievable target — that's roughly the level where AI-cited awareness starts compounding into pipeline. New brands typically start near zero. The curve depends on content quality, entity signals and third-party authority, which is exactly what the 90-day playbook in your workspace sequences.

llms.txt is a proposed standard (Jeremy Howard, 2024): a markdown file at your site root that gives AI crawlers a curated summary of who you are and where your important content lives. Honestly: no major AI provider has formally committed to honouring it yet. But implementation cost is near zero, adoption is spreading, and not having one is starting to look like a tell. Cited generates yours from your workspace facts — and yes, this site publishes its own.

Published verbatim without review, yes — and fabricated statistics are worse: LLMs cross-reference, and an invented claim that contradicts authoritative sources can get your page excluded from citations entirely. The rule we build into every Cited tool: AI drafts, a human verifies, and every specific claim must be true. If a human couldn't put their name to it, don't publish it.

No. Cited works entirely on public signals: your published pages, the prompts real customers ask AI assistants, and the answers those assistants give. Nothing in your workspace touches customer records, and the monitoring agent has no access to your inbox, CRM or application data — a boundary we consider non-negotiable.