AI search glossary

Terms

AI site search

AI site search is a search feature that understands a visitor's question in natural language and returns a written answer drawn from a single website's own content, rather than a list of blue links. It reads the relevant pages for the visitor and responds directly. Achla AI adds this to any site through one embedded script.

Answer engine

An answer engine is a search system that responds to a query with a direct, composed answer instead of a ranked list of documents to open and read. It interprets intent, gathers the relevant material, and writes a concise reply. Achla is answer-first by design, so visitors get the response rather than the homework.

Generative answers

Generative answers are search responses written on the fly by a large language model, phrased in natural sentences rather than pulled verbatim from one page. The model composes a reply from the material it is given. Ungrounded generation can hallucinate, which is why Achla constrains every generative answer to your indexed content.

Grounded answers

A grounded answer is a generated answer tied to specific source passages, so each claim can be traced back and verified rather than taken on trust. Grounding is what separates a reliable answer engine from a chatbot that guesses. Achla grounds every answer in your indexed pages and shows the sources it used.

Citations / sources

Citations are the references an answer engine attaches to a reply, pointing to the exact pages the answer was built from so a reader can confirm it. They turn an opaque answer into a checkable one. Achla treats citations as mandatory: if it cannot cite a source, it declines to answer instead of inventing one.

RAG (retrieval-augmented generation)

RAG is a technique that first retrieves relevant passages from a knowledge source, then feeds them to a language model so its answer is generated from that specific material rather than from memory alone. It keeps answers current and grounded. Achla uses a retrieval-then-generate pipeline over each site's crawled content.

Semantic search

Semantic search is retrieval that matches on meaning rather than exact words, so a query for "return policy" can surface a page titled "refunds and exchanges." It compares the intent behind text, not just the characters. This lets an answer engine find the right passage even when the visitor's phrasing differs from the page's.

Keyword search

Keyword search is the classic approach that matches the literal words in a query against the words on a page, ranking results by term frequency and similar signals. It is fast and predictable but misses synonyms and paraphrases. Modern AI site search blends keyword matching with semantic understanding to catch both exact and meaning-based matches.

Embeddings

Embeddings are numeric representations of text that place similar meanings close together in a mathematical space, letting a system measure how related two pieces of text are. They are the engine behind semantic search and RAG retrieval. An answer engine embeds both your content and each query to find the passages that best match intent.

Ticket deflection

Ticket deflection is the reduction in support requests that happens when visitors find answers themselves instead of contacting a team. A good self-serve answer engine resolves repeat questions at the point of need. By answering directly from your documentation and FAQs, Achla helps visitors resolve common questions without opening a ticket.

Site search widget

A site search widget is an embeddable, self-contained search interface a site owner drops onto their pages, usually with a single script tag and no backend to build. It handles the input box, the results, and the styling. Achla's widget runs in bubble, inline, or attached-input modes and needs only one line of code.

Multilingual search

Multilingual search is the ability to understand a query and reply in the visitor's own language, even when the underlying content is written in another. It matches meaning across languages rather than requiring a match in the same one. Achla answers in the language a visitor asks in, grounded in your content whatever language it is stored in.

GEO (generative engine optimization)

GEO is the practice of structuring and phrasing web content so that AI answer engines — such as AI Overviews, ChatGPT, and Perplexity — can quote it accurately when they respond to users. Clear, self-contained definition blocks are especially citable. This glossary is written in exactly that quotable, front-loaded style.