The new standard for local visibility

Just as robots.txt guides traditional search crawlers, these markdown files guide generative AI. llms.txt is a concise routing document: it tells an assistant what your business is and where the key content lives. llms-full.txt is the comprehensive corpus — your services, your facts, your answers — for models that read the whole thing.

Search Engine Optimisation convinces a search engine to list your page as a link. Generative Engine Optimisation structures your facts so an AI assistant can find, trust and repeat them in its answer. The first earns a click; the second earns a citation.

Anyone can paste a file. The work is in the server: caching that serves the right file, headers that let the crawlers in, structured data that agrees with what the page says, and a way to verify all of it live. When that is done wrong the models read nothing, and nobody notices for months.

Google has shown agents that browse, select and check out across vendors on a buyer’s behalf. To be in that loop your products, prices and policies must be machine-readable and your site must answer fast. The packages put the data layer in place; the speed engineering makes it answer.

No. This is a rapid architectural patch. We do not alter your design, your branding or your front end; we work in the machine-readable layer and verify each change live before we move on.

A server-side channel that lets an AI model read your data directly — a direct pipeline rather than a scrape of your pages. Most local businesses need the files and the entity graph first; WebMCP readiness is prepared in the Upgrade tier and built in full in a Strategic Web Architecture rebuild (from £3,500) or a GEO retainer.

Because regulated sectors carry data-governance mandates — the Data Security and Protection Toolkit for UK healthcare, the SRA’s rules for solicitors — and a site that opens itself to AI crawlers must also harden its perimeter: a Grade A/A+ security rating, a content security policy engineered for your site, and alignment with the frameworks your regulator expects.

Faster than traditional SEO, which can take six to twelve months. The technical changes show on the very next scan; generative models refresh their context on their own schedules, so changes in what they say, when they come, usually show within weeks. We track it weekly so you see whether the line moves rather than take our word for it — and no one can guarantee that a model will cite a business.