Elasticsearch Relevance Engine Quick Start

In this 3-step Quick Start series, you'll learn about the Elasticsearch Relevance Engine™ (ESRE), designed to power AI search applications. See ESRE's features in action such as enabling semantic search in one-click, hosting and using your own custom ML model, and integrating with LLMs to build generative AI experiences.

Video thumbnail

Now it's your turn

Now that you've watched the Quick Start videos, follow the steps below to practice what you've learned. If you don't have an Elastic Cloud instance, spin up a 14-day free trial.

  • Step 1

    1. Learn about the Elasticsearch Relevance Engine features
    2. Create a "Quick Start: Elasticsearch Relevance Engine" deployment
    3. Launch Elasticsearch Relevance Engine
  • Step 2

    1. Ingest data using Elastic's native MySQL connector
    2. Enrich data with Elastic's proprietary machine learning model (ELSER)
    3. Try full text search and semantic search
    4. Upload a third-party ML Model from Hugging Face
    5. Generate vector embeddings
    6. Try vector search and hybrid search
  • Step 3

    1. Open a Google colab notebook example from Elasticsearch-labs to build a Generative AI Application
    2. Ingest a Wikipedia dataset with OpenAI embeddings in Elasticsearch
    3. Build a simple Streamlit application to implement Retrieval Augmented Generation (RAG) using the Elasticsearch Relevance Engine
    4. Ask a simple question, like "Who is Beethoven?" See the relevant documents retrieved from Elasticsearch and pass them via context window to the transformer model

See recent posts on AI search, LLMs, and vector databases at Elastic Search Labs.