Integrations

# AI search setup

Pick and look after the model behind semantic search.

-   Choose an embedding model and check whether AI search is on.
-   When the embedding configuration changes, follow the recalculation as it runs.
-   The screen tells apart active, unconfigured and failed recalculation states.

AI search needs an embedding model that's turned on, and the index has to be recalculated.

![The AI Search page shows the embedding model selector. No embeddings were generated.](https://welley.com/assets/media/admin-demo-integration-ai-search-2000.152b4e2c.webp)

The AI Search page shows the picker for the embedding model. We didn't generate any embeddings.

[Previous: MCP servers](https://welley.com/integrations/integration-mcp/) [Next: Webhooks](https://welley.com/integrations/integration-webhooks/) [Back to Integrations](https://welley.com/integrations/#server-integrations)