Skip to main content
Integrations let LightOn pull from systems where your documents already live, instead of asking you to upload them by hand. Set one up once and the workspace stays in sync: additions, updates, and deletions in the source flow through automatically on the next import.

​
Datasources

Datasources sync a workspace to an external file store. The workspace becomes read-only on LightOn’s side; the source system remains the single source of truth.

Google Drive

Sync a workspace to a Google Drive folder via a service account.

Microsoft SharePoint

Sync a workspace to a SharePoint site using an Azure app registration.
Both datasources can be configured from the console UI (create a new workspace, then open it and choose to sync it from inside the workspace’s settings) or via the API by issuing PATCH /api/v3/company/workspaces/{id} with a datasource object. See each datasource’s page for the credentials and filter criteria it expects.

​
LLM Providers

LLM provider guides show how to combine LightOn retrieval with a hosted model to build RAG pipelines. LightOn handles search and ranking; the provider handles generation.

Lyceum Serverless Inference

Use LightOn search as the retrieval layer for vision-language and other models hosted on Lyceum.

Scaleway Generative APIs

Use LightOn search as the retrieval layer for Llama, Mistral, and other models hosted on Scaleway.

OVHcloud AI Endpoints

Use LightOn search as the retrieval layer for Qwen, Mistral, Llama, and other models hosted on OVHcloud.

Inceptron

Use LightOn search as the retrieval layer for Llama, Kimi, MiniMax, and other models hosted on Inceptron.