Sanity Context and content APIs (opens in a new tab)
The official reference for the tools used in this lesson.
Link
Keeping the index fresh
Content changes and stale chunks keep getting retrieved. Incremental reindexing driven by content events beats rebuilding everything nightly.
Lesson 2.35m 58sAdvanced6.8k students
Lesson length
5:58What this lesson covers
- Reindex incrementally on content change events
- Delete chunks whose source no longer exists
- Version the index so a bad run can be rolled back
This lesson sits in Ingestion and Indexing, part of Retrieval-Augmented Generation from Scratch. It assumes what came before it and leads directly into the next lesson in the module.
In this lesson you will
- Reindex incrementally on content change events
- Delete chunks whose source no longer exists
- Version the index so a bad run can be rolled back
Resources
Personal notes are coming soon.