Skip to content

Vector Demos

Vector Search and GenAI, a curated collection of demo notebooks and apps.

Notebooks

This is a collection of quickstarts and tutorials, available either as stand-alone Jupyter or Google Colab notebooks (usually both), and involving Astra DB and generative AI -- specifically, vector-search-powered use cases.

Overview Prerequisites Links
Learn about using Vector Search to find content related to a query, and then pass that to an LLM to understand how the RAG pattern works for AI powered chatbots. You will need an Astra account with a Serverless Cassandra with Vector Search database. Moreover, an OpenAI API Key is required. Open in Colab Or, download the notebook.
Learn how to use Vector Similarity Search to find images based on natural language descriptions using CLIP. You will need an Astra account with a Serverless Cassandra with Vector Search database. Open in Colab Or, download the notebook.
Try a simple Question-Answering demo powered by a vector-capable database instance. You will use the Astra integration for LangChain and OpenAI for the embeddings and the LLM (Large-Language-Model). You will need an Astra account with a Serverless Cassandra with Vector Search database. Moreover, an OpenAI API Key is required. Open in Colab Or, download the notebook.
Understand Vector Search and the RAG pattern by building a simple generator of "philosophical quotes" which uses Astra for storage and OpenAI for retrieval+generation. You will need an Astra account with a Serverless Cassandra with Vector Search database. Moreover, an OpenAI API Key is required. Notebooks (including Colab links) hosted by openai-cookbook.

If you open the notebooks in Colab, and would like to make changes to them, choose "Save a copy in Drive" from the File menu in Colab.

Sample applications

Check this section for full applications making use of Vector Search, Astra DB and other GenAI technologies.

  • AI-powered "Hotel search demo" (uses Vector Search and GenAI personalization; has a one-click "open in Gitpod" button)
  • AI-powered "FLARE QA on PDF files" application (vector-search-based question-answering client+API setup; has a one-click "open in Gitpod" button)
Last update: 2023-10-13