Skip to main content

Quick Start

This page describes how to get started quickly with AVS by using the product preview.


  • Access to an AVS preview environment. Request access at
  • A local Docker installation

1. Run the Quote Search Example

In the following example, replace <AVS-HOST-IP> with the IP address of your AVS host in the preview environment.

docker run -e AVS_HOST="<AVS-HOST-IP>" \
-p 8080:8080 \

After a few moments, go to http://localhost:8080/search to perform a search. Try using natural language searches, such as "Tell me about whales?".

2. Index your Personal Photos

This example illustrates how to index photos for semantic search. While it takes a bit longer to run, it gives you a better of idea of how to leverage open source models to build an application. This example uses the open source CLIP model to generate image embeddings by processing any JPEG or PNG files in the folder specified or any of its sub-folders.

docker run -e AVS_HOST="<AVS-HOST-IP>" \
-v ~/Pictures:/prism/static/images/data \
-p 9090:8080 \

Running this example locally generates local vector embeddings from your images, but these images will not leave your computer. Only vector embeddings are sent to AVS. This process can take several minutes depending on the number and size of your photos.

After a few moments, go to http://localhost:9090/search to perform a search.

3. Try the LangChain Python Notebook

Now that you've seen a few example apps, try walking through our Python Notebook for LangChain. The notebook walks you through building a semantic search example for a set of quotes.

Next Steps

After walking through the LangChain example, dive deeper by checking out our examples and learning about our Python Client.