The words, defined plainly.
AI search has collected a lot of acronyms in a short time, most of them explained badly. Here is what each one actually means, in one or two sentences, with no pitch attached.
SEO is getting a page to rank in a search engine's list of links. AEO is structuring content so an AI assistant can quote it. GEO is becoming a business an assistant will vouch for when someone asks who to hire. DaxReach runs all three as one system and calls it the GAS engine.
SEO(Search Engine Optimisation)
SEO is the work of getting a page to rank in a search engine's ordinary list of results, so that someone searching for what you sell finds your site among the links.
For a local service business it is mostly unglamorous: a page for each service and each area served, written in the words customers actually use; a site structure a crawler can follow; and enough third-party evidence that the business is real. It is the oldest of the three disciplines and still where most search volume sits.
See also: AEO, GEO, GAS engine
AEO(Answer Engine Optimisation)
AEO is the work of structuring content so an AI assistant can lift a clean, accurate passage from it and quote that passage in an answer, instead of paraphrasing a competitor who made it easier.
In practice it means answering the question in the first sentence rather than the fourth paragraph, keeping each answer self-contained enough to survive being excerpted, and being unambiguous about who the answer is about. Ranking and being quoted are different jobs: a page can sit at position one and still be skipped by an assistant that could not extract a usable sentence from it.
See also: SEO, GEO, Answer engine
GEO(Generative Engine Optimisation)
GEO is the work of becoming a business an AI assistant is willing to vouch for when someone asks who to hire — which depends less on your own website than on what independent sources say about you.
An assistant asked to recommend somebody is making a judgement about risk, and it resolves that from corroboration: a complete and consistent business profile, listings that agree with each other, genuine reviews, and mentions on sources it already trusts. This is the part that cannot be fixed on your own site, and the part that takes longest.
See also: AEO, Entity, AI citation
GAS engine(Generative, Answer, Search)
GAS is DaxReach's name for running Generative, Answer and Search engine optimisation as one system rather than three separate projects — the three acronyms in the order that spells the word.
The argument for treating them as one is that they share most of their inputs and disagree about none of them: the same clean structure that helps a page rank is what makes it quotable, and the same consistent listings that reassure a search engine are what reassure an assistant. Splitting them across vendors mostly produces three people doing the overlapping 70% three times.
Answer engine
An answer engine is a system that responds to a question with a written answer rather than a list of links — ChatGPT, Perplexity, Claude, Google's AI Mode and AI Overviews all behave this way.
The consequence for a business is that the result page stops being a list it can appear somewhere on, and becomes a short answer it is either named in or absent from. Short lists exclude, and being good at the work is not what puts you on one.
See also: AEO, AI citation
AI citation
An AI citation is an instance of an assistant naming a business, or linking its page, inside an answer it gives to a user's question.
Citations are the unit that matters for this kind of visibility, and they are reported per engine rather than averaged, because the engines disagree with each other often enough that a blended score hides the useful information. A citation is not a ranking: there is no position two.
See also: Answer engine, GEO
Entity
An entity is the record a search engine or AI model holds about a thing that exists — a company, a person, a place — assembled from every source that mentions it, and kept separate from the words used to describe it.
A business with a thin or contradictory entity is hard to recommend, because there is nothing to check the claims against. Making the name, address, categories, hours and descriptions agree everywhere they appear is the cheapest available improvement to it, and inconsistency is one of the most common reasons a business is quietly left out of an answer.
See also: GEO
llms.txt
llms.txt is a plain-text file at the root of a website that summarises what the site is and links its most useful pages, written for AI models rather than for people or search crawlers.
It is an emerging convention rather than a standard any engine has committed to honouring, and it should be treated that way: cheap to publish, honest about what the site actually contains, and not a substitute for the pages themselves. DaxReach publishes one at /llms.txt.
See also: AEO
Human in the loop
Human in the loop means an AI system's output is reviewed and approved by a person before it takes effect, rather than being published automatically.
For marketing work the distinction is whether anyone read the page before it went on your site under your name. Agents can do the volume — research, drafting, structuring, monitoring — while approval stays a human decision, which is the arrangement DaxReach uses and says so on every page.
Service as a software
Service as a software is a model where the customer buys a finished outcome that the provider delivers using software, rather than buying a software licence and having to operate it themselves.
It inverts the usual SaaS arrangement: with SaaS the licence is the product and the work remains yours, whereas here the work is the product and the software is how the provider does it at a price a purely manual service could not reach.
See also: GAS engine
Want this done for your business rather than explained?
The four services, what each includes and who each is for — no figures hidden behind a form.