Скачать или смотреть How to Run Hybrid Search with Reciprocal Rank Fusion (RRF) | Spice.ai Demo

In this tutorial, Ellison from Spice demonstrates how to perform hybrid search on real-time data using the Reciprocal Rank Fusion (RRF) algorithm.

Combining results from multiple search methods like keyword and vector search historically has required complex pipelines and custom ranking logic. RRF solves this by merging results into a single, high-quality ranking.

With Spice, you can do it all in one query.

💡 In this demo, you’ll learn how to:

Capture live Bluesky posts in real time
Automatically index and embed them
Run hybrid search with RRF across text and vectors
Apply ranking weights and recency boosts
Tune RRF parameters for optimal results
Search across multiple languages
Enable agentic support to query insights with OpenAI

All of this happens inside a single runtime. No ETL, separate search engines, or custom ranking logic required.

📘 Learn more:

RFF cookbook: https://github.com/spiceai/cookbook/b...
RFF docs: https://spiceai.org/docs/reference/sq...

💬 Questions? Reach out to us on Discord:   / discord