All projects · AI Agents & RAG
KB Ingestion — OpenAI + Pinecone
A production-grade RAG ingestion pipeline orchestrating 14 clean atomic knowledge chunks for Ryobi Coffee Co., vectorizing them via OpenAI…
A production-grade RAG ingestion pipeline orchestrating 14 clean atomic knowledge chunks for Ryobi Coffee Co., vectorizing them via OpenAI text-embedding-3-small (1536 dimensions), and upserting into Pinecone with rich metadata in a sequential batch loop.
- 14 clean, atomic, self-contained knowledge chunks ensuring pinpoint semantic retrieval without dilution
- Split-In-Batches sequential loop managing item-by-item ingestion with zero rate-limit spikes
- OpenAI text-embedding-3-small generating dense 1536-dimensional float vector representations
- Pinecone index namespace isolation ('ryobi') with structured metadata attachment (title, category, text, source)
Built with: n8n Manual Trigger, Set node (config constants), JavaScript Code Node (14 atomic chunks), Split In Batches loop (batchSize 1), OpenAI Embeddings API (text-embedding-3-small, 1536-dim), Pinecone Vector DB (vectors/upsert via HTTP Request)