## Key features

- **Chat search**: Describe your ideal stay in plain language (e.g., “Find me a family-friendly beach house in Europe under $300/night”) and get relevant property results with matching keywords highlighted.
- **Iterative refinement**: Narrow results through follow-up messages. Start broad, then refine: “Only alpine cabins” then “for 8 persons” to progressively filter down to the perfect match.
- **Classic search**: Switch between chat and traditional keyword search with filters for a familiar booking experience.
- **Meilisearch chat route**: Uses the `/chat` route with ChatGPT 5.2 as the LLM and [Zembed](https://www.zeroentropy.dev/) embeddings via HuggingFace Inference Endpoints for semantic understanding.
- **Built with Lovable**: The frontend is entirely generated with Lovable, demonstrating how quickly you can prototype a full search experience on top of Meilisearch.

This demo has minimal guardrails. The chat interface may occasionally produce unexpected responses.

## Example conversation

1. **“Find me a family-friendly beach house in Europe under $300/night”** - Returns 6 beach properties across Italy, Spain, and Greece, all under budget, with matching terms highlighted.
2. **“I would like only alpine cabins”** - Filters down to 4 cabin-style properties in alpine regions.
3. **“For 8 persons”** - Narrows to a single property that fits all criteria.

## Links

[**Try the demo** 
\
Search properties with chat or classic search](https://lodging-lark.lovable.app/)

[**Chat route docs** 
\
Learn about Meilisearch conversational search](/content/docs/capabilities/conversational_search/overview/index.html)