Grep your lakehouse: Search-first retrieval for DuckDB-powered agents
summary
Extends Duck Lake/DuckDB by adding index-backed full‑text and semantic retrieval as SQL extensions, using TenTen TV indexes stored beside Parquet files and Postgres catalog metadata; implements CREATE INDEX that returns row IDs and scores as virtual columns, adds an @@ operator for text and a vector similarity operator, and exposes a star column to search across all fields. Pipeline syncing search and DB is removed and queries are rewritten to push scans to DuckDB efficiently.
AI-written summary · about
chapters (6)
- 0:00Search-first retrieval intro
- 1:50Speaker background
- 3:38Search engines and indices
- 5:43DuckDB search extension
- 9:19Duck Search implementation
- 11:48Call to action
related talks
06mDuckDB Applications in Information Retrieval
21mWhat can Postgres learn from DuckDB?
38mFrom Big Data to DuckDB: Ryan Boyd (MotherDuck) on Hyper-Tenancy, Duck Lake, and Making Data Fun.
40mDuckDB Meets Data Lakes
25mA quack of all trades: DuckDB, a versatile analytical database to keep in your toolkit
01h 02mAnalytics for not-so-big data with DuckDB
47mWhy DuckDB could be the engine for your Data Lake?
19mIn-Process Analytical Data Management with DuckDB