# Add search

You sometimes want visitors to be able to find the product or article they want on your site. Search is not a part you bolt on afterward. It starts when you build the data structure, by deciding **which fields can be searched and in what way**. **Tell the agent what you want people to search by, and it sets up the fields to match.**

## Tell your AI like this {#tell-your-ai}

> Connect this book search to WEEGLOO. Let people search by title, and also by a word inside the plot summary.

## What the AI then does {#what-the-agent-does}

It chooses the field type to match how you want to search, and connects the search box to it.

- Clear-cut values like a title are set up to be found by an exact match or by the first few characters.
- Long text like a plot summary gets full-text search turned on, so the book can be found by a word inside the text.
- Advanced Search, which applies several conditions at once, is also available.

Search works most smoothly when you decide it together with the data structure. Adding search to a structure you already built is covered in [Content modeling](/getting-started/core-concepts/content/content-modeling.md).

## Check that it worked {#how-to-check}

If you type part of a title into the search box and only that book shows, and you can also find it by a word inside the plot summary, then it is done.

![Add search](https://youtu.be/9F3jb57tiwA)

## What to do next {#what-to-do-next}

- [Content modeling](/getting-started/core-concepts/content/content-modeling.md): covers how the way each field type is searched differs.
- [API reference](/api/reference/common/query-parameters.md): covers technical specifications such as the request format for applying search and filter conditions.