Load Confluence pages into Pinecone for semantic search

Setup steps

1. 1

   Choose the project and embedding model

   Open Settings → Embedding Model in the project you will use. Select a provider and model, save its API key, and test the connection. Record the model dimension before configuring Pinecone.

   Checkpoint: The embedding connection succeeds and you know the configured vector dimension.
2. 2

   Configure Pinecone for the same project

   Open Settings → Vector Store, select Pinecone, enter the index name and optional namespace, then save the Pinecone API key. Test the connection. The Pinecone index dimension must equal the embedding dimension.

   Checkpoint: Vector Store shows a successful Pinecone connection and the expected index dimension.
3. 3

   Connect Confluence Cloud

   Open Sources, add Confluence, and complete the Atlassian consent flow. Choose an Atlassian site where your account can read the content you need. Atlassian permissions still determine which spaces and pages are visible.

   Checkpoint: The Confluence source appears as connected without asking you to paste an Atlassian token.
4. 4

   Browse and select pages

   Open Browse on the source. Expand a space to load its root pages and folders, then expand only the branches you need. Select the current pages you want and start the sync.

   Checkpoint: The sync is created for the selected project and the chosen pages appear in Documents.
5. 5

   Verify retrieval before using the data

   Wait for the documents to show a synced status, then open Playground and search for a phrase that appears in one of the pages. Inspect the returned source title and relevance score.

   Checkpoint: A known Confluence page appears in the results for a specific test query.