Guides
Why built-in AI search matters for documentation
Readers ask questions in plain language. Docfiy indexes your content so answers surface without extra setup.
Keyword search is not enough
Traditional docs search only matches keywords. Readers type “how do I rotate API keys” and get zero results because your page title says “Credential lifecycle”.
Semantic search closes that gap by matching intent, not just tokens — so the right section surfaces even when wording differs.
One index, two surfaces
Docfiy builds a single content index when you publish. The same embeddings power sitewide search and the floating AI assistant — grounded in your pages, not generic model knowledge.
| Capability | Keyword search | Docfiy AI search |
|---|---|---|
| Matches paraphrases | Limited | Yes |
| Setup required | Often manual tuning | Included on publish |
| Assistant grounding | Separate product | Same index |
| Maintenance | Re-index pipelines | Updates on recrawl/publish |
What to write for better answers
Clear H2 headings, short paragraphs, and task-oriented titles improve both human scanning and retrieval quality. Put the answer in the first paragraph when possible — assistants and search snippets prefer direct openings.
We cut “where is this documented?” tickets by pointing search at Docfiy instead of Slack threads.
Frequently asked questions
Is AI search included on the free plan?
Search is available on paid plans. Free plans include published pages and editing; upgrade when you need search and assistant features for readers.
Does the assistant hallucinate?
The assistant is constrained to your published documentation index. It should cite or paraphrase your content rather than invent product behavior.