Glossary
AI search engine search that writes the answer.
An AI search engine answers a question in written form, built from web pages it retrieves and cites. How they work and how Prefer tracks your brand in them.
AI search engine An AI search engine answers a question with a written response built from web pages it retrieves and cites, instead of a list of links. Prefer tracks how five of them answer about your brand.
An AI search engine answers a question with a written response built from web pages it retrieves, usually with citations, instead of returning a ranked list of links. Prefer tracks how five of them answer about your brand: ChatGPT, Gemini, Perplexity, Google AI Overviews and Google AI Mode. Claude is tracked on Enterprise.
How an AI search engine works#
Most follow the same four steps, a pattern known as retrieval-augmented generation:
- Interpret the question. The model works out what is being asked and whether it needs the web at all.
- Search. It runs one or more queries. Google documents that AI Overviews and AI Mode may use “query fan-out,” issuing several related searches across subtopics. See query fan-out.
- Read. It pulls passages from the pages that come back.
- Write and cite. It writes the answer from those passages and links the sources.
Each engine documents its own version. Google’s Grounding with Google Search describes Gemini running searches and returning answers with inline citations. OpenAI’s crawler documentation says OAI-SearchBot is used “to surface websites in search results in ChatGPT’s search features.” Perplexity’s bot documentation says PerplexityBot exists to “surface and link websites in search results on Perplexity.”
The main types#
- Chat assistants with search. ChatGPT, Gemini and Claude answer in a conversation and search the web when the question needs it. See does ChatGPT use live search.
- Answer-first search engines. Perplexity leads with a cited answer. In Prefer’s September 2026 test it showed sources on every answer (study).
- AI layers inside classic search. Google’s AI Overviews sit above the normal results, and AI Mode is a separate conversational tab.
For a ranked comparison of each, see the best AI search engines.
Why it matters for AI visibility#
In a classic search engine you can rank tenth and still get a click. In an AI search engine the result is a paragraph that names a few brands and cites a handful of pages. In Prefer’s September 2026 run, web-grounded ChatGPT answers cited 3.8 sources on average (how many sources ChatGPT cites). You are named, cited, or absent.
The engines also disagree. In Prefer’s September 2026 test of 960 answers, of 1,329 sites the four chat engines cited, 966 were cited by only one engine, while 23 of the 25 most-cited sites were cited by three or four (study data). Winning on one engine says little about the others, so results should be read per engine.
Example#
A buyer asks “what is the best project management tool for a 20-person agency?” A classic engine returns ten links. An AI search engine searches several variations, reads review sites, community threads and vendor pages, and writes a short list of four tools with reasons and citations. If your product is not on the pages it read, it is not in the answer.
Common confusions#
- AI search engine vs chatbot. A chatbot answering only from training data is not searching. It becomes a search engine when it retrieves live pages and cites them.
- AI search engine vs answer engine. The terms overlap heavily. “Answer engine” stresses the single answer; “AI search engine” stresses the retrieval behind it.
- Being mentioned vs being cited. A brand can be named in the answer without its site being a cited source, and the reverse. Both count, and they are measured separately.
How Prefer helps#
Prefer runs your buyer prompts on each engine, shows whether you are named, which sources each answer cites, and your share of voice against named competitors. Check where you stand now with the free AI Visibility Checker, and make sure the search crawlers can reach you with the Crawler Access Checker.
In context
The term in a sentence.
Related questions
People also ask.
- What is the difference between an AI search engine and Google?
- Which AI search engines matter most for brands?
- How do AI search engines choose which sites to cite?
Questions
Asked plainly.
What is an AI search engine in simple terms?
It is a search tool that reads the web for you and writes the answer, with links to the pages it used. Prefer tracks five of them for your brand: ChatGPT, Gemini, Perplexity, Google AI Overviews and Google AI Mode, with Claude on Enterprise.
Is Google an AI search engine?
Partly: Google Search now shows AI Overviews on many queries and offers AI Mode as a full AI search experience, alongside the classic links. Prefer tracks both Google surfaces separately, because the same brand can be cited in one and missing from the other.
How do AI search engines decide which sites to cite?
They run one or more searches, read the pages that come back, and cite the ones they use in the answer. Prefer shows the sources behind each answer for your prompts, so you can see which pages each engine actually reads in your category.
Do all AI search engines cite the same sites?
No. Prefer's September 2026 test found that of 1,329 sites the four chat engines cited, 966 were cited by only one of them, though the biggest sites were shared. That is why Prefer reports results per engine rather than as one blended score.
Keep reading