Skip to main content
Output and upsert components write pipeline data out to an external destination, such as a relational database, a SaaS platform, or a vector database. Where the destination supports it, they can upsert records, inserting new rows and updating existing ones in a single step, rather than only appending data.

​
Use cases

Output and upsert components are especially useful when a pipeline needs to push processed or enriched data to a downstream system, rather than only loading it into the warehouse. Use output and upsert components to:
  • Write pipeline results back to an operational or SaaS system, keeping it in sync with your data warehouse.
  • Bulk load data into a relational database as the final step of a pipeline.
  • Insert or update records in place, avoiding duplicate rows when a pipeline runs repeatedly against the same destination.
  • Write embeddings and associated metadata into a vector database to support semantic search or retrieval-augmented generation (RAG).