> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scentxp.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Capabilities

## What ScentBot can do

ScentBot is more than a chatbot — it is an AI fragrance specialist trained on WikiParfum's database of 31,000+ perfumes. This page describes the capabilities available to users during a conversation.

## Natural language search

Users describe what they want in their own words, and ScentBot translates that into structured fragrance queries. Supported search dimensions include:

| Dimension                 | Examples                                                  |
| ------------------------- | --------------------------------------------------------- |
| **Keywords**              | "a fresh summer perfume", "something elegant for evening" |
| **Ingredients and notes** | "with bergamot and sandalwood", "no patchouli"            |
| **Fragrance families**    | "oriental", "floral", "woody"                             |
| **Brand**                 | "show me Chanel perfumes", "anything by Tom Ford"         |
| **Perfumer**              | "fragrances by Francis Kurkdjian"                         |
| **Gender**                | "for men", "unisex options"                               |
| **Intensity**             | "something light", "a strong, long-lasting scent"         |
| **Price tier**            | niche, prestige, or mass classification                   |
| **Release year**          | "launched in 2023", "perfumes from the 90s"               |
| **Fragrance type**        | Eau de Parfum, Eau de Toilette, Extrait, Eau Fraiche      |

These dimensions can be combined in a single query. ScentBot understands context and refines results as the conversation progresses.

## Image-based search

Users can upload a photo — an outfit, a look, or a setting — and ScentBot suggests fragrances that match the visual style. This is useful for customers who know the aesthetic they want but cannot describe it in fragrance terms.

## Gift recommendations

ScentBot guides users through gift selection with tailored suggestions based on the recipient's preferences:

* **Safe choice** — recommendations based on known preferences
* **Surprising choice** — niche or unexpected suggestions for adventurous recipients
* **Open-ended** — broad recommendations when the user has no specific direction

## Fragrance comparison and details

Users can ask ScentBot to compare two or more fragrances, explain a perfume's composition, or describe what a fragrance smells like. ScentBot provides ingredient breakdowns, family classification, and perfumer information.

## Similarity and dupes

ScentBot can find fragrances similar to a given perfume — useful for customers looking for alternatives or more affordable options within your catalog.

## Layering suggestions

For customers interested in fragrance layering, ScentBot can suggest complementary pairings — recommending fragrances that combine well with a selected scent.

## Multi-language support

ScentBot supports **25+ languages**. Set the language via the [`language` attribute](/scentbot/configuration) using an ISO 639-1 code. The entire conversation — including recommendations, ingredient names, and UI text — adapts to the selected language. Arabic (`ar`) enables right-to-left layout.

## Session persistence

Conversation history is stored in the browser and persists across page navigations within the same site. Users can leave a product page, browse other sections, and return to find their conversation intact.

The history expires after a configurable period (default: 12 hours) with no user interaction. See the [`chat-history-expiration-hours` attribute](/scentbot/configuration) to adjust this.

## Catalog-scoped results

All recommendations and search results are scoped to the products in your [Product Feed](/catalog-integration/product-feed). Users only see fragrances they can purchase from your store.

## Brand customization

ScentBot adapts to your brand identity:

* **Tone of voice** — conversational style matches your brand personality
* **Welcome message** — customizable greeting when the widget opens
* **Quick-reply FAQs** — predefined questions displayed as suggestion chips
* **Visual styling** — fonts, colors, and layout via [custom CSS](/scentbot/configuration#appearance-and-customization)

<Note>
  Tone of voice, welcome message, and FAQ configuration are managed by ScentXP during onboarding. Contact your account manager to customize these settings.
</Note>

## Next steps

<CardGroup cols={2}>
  <Card title="Installation" icon="download" href="/scentbot/installation">
    Get ScentBot running on your website.
  </Card>

  <Card title="Configuration" icon="gear" href="/scentbot/configuration">
    All widget attributes for customizing behavior.
  </Card>
</CardGroup>