Sync WordPress and WooCommerce content with Pinecone using OpenAI embeddings

Quick overview

This workflow syncs WordPress/WooCommerce pages and blog posts into a Pinecone vector index by fetching content in small batches, generating OpenAI embeddings, and tracking indexed items in an n8n Data Table to handle updates and deletions.

How it works

1. Runs on a manual trigger or on a daily cron schedule.
2. Fetches up to 100 published WordPress pages and up to 100 published posts (IDs and modified timestamps) via the WooCommerce/WordPress REST API.
3. Compares the live IDs/modified dates with the indexed_wc_blog_v1 n8n Data Table and selects up to staticBatchSize pages/posts that are new or changed.
4. Fetches the full content for the selected batch from the WordPress pages and posts endpoints and combines the results.
5. For each page/post, strips HTML, prepares metadata, deletes any existing Pinecone vectors for the same fileId, then splits text, generates OpenAI embeddings, and inserts the chunks into Pinecone.
6. Upserts the page/post’s fileId and dateModified into the indexed_wc_blog_v1 Data Table to record what was indexed.
7. Identifies entries that exist in indexed_wc_blog_v1 but no longer exist in WordPress and deletes their vectors from Pinecone and their rows from the Data Table.

Setup

1. Add WooCommerce API credentials with access to the WordPress REST API endpoints for pages and posts.
2. Add an OpenAI API credential for creating embeddings.
3. Add a Pinecone API credential, create or select a Pinecone index, and set the index host and namespace values.
4. Create an n8n Data Table named indexed_wc_blog_v1 with text columns productId, indexedAt, and dateModified.
5. Update wooBaseUrl, pineconeIndexName, pineconeIndexHost, pineconeNamespace, and staticBatchSize in the “WC: Settings (static content)” node before enabling the schedule.