Create a store

By the end of this page you will have created a Content Store, ready to fill with documents or crawled pages.

Prerequisites

Admin or Developer role.
At least one embedding configuration in the project (set up under Configuration → Embeddings). The create dialog needs one to embed your content.

Create the store

Open the create dialog

On the Content Store page, click Create Store. The Create RAG Store dialog opens.

Name the store

Enter a Store Name (for example “Product Documentation”) and an optional Description.

Create RAG Store dialog with store name, description, embedding model, vector schema, and chunking settings

Choose an embedding model

Pick an Embedding Model from your project’s embedding configurations. If the dropdown shows “No embedding configurations found”, add one under Configuration → Embeddings first.

Choose a vector storage schema

Select the Vector Storage Schema — Shared by default, or Dedicated on Pro+ plans.

Set chunking

Under Chunking Settings, choose a Chunking Strategy and adjust the chunk size and overlap:

Fixed Size — split by character count. Fast and consistent.
Sentence-Based — split at sentence boundaries. Better context.
Semantic — split by meaning. Best quality, slower.

💡

The defaults (Fixed Size, chunk size 1000, overlap 200) are a solid starting point. Larger chunks keep more context per result; more overlap reduces the chance of splitting an idea across chunks.

Create

Click Create Store. It appears in the list, ready for content.

⚠️

You choose the embedding model and chunking once, at creation — they can’t be changed later. To use different settings, create a new store.

Common issues

“No embedding configurations found” — the store needs an embedding model. Add an embedding configuration under Configuration → Embeddings, then reopen the dialog.
No Dedicated option — dedicated vector schemas are a Pro+ feature; the shared schema is used otherwise.

Add documents — link content from your Documents store.
Web crawler — index a website.