Today we released the Astra Assistants API, a service that is API-compatible with OpenAI's new Assistants API. Users can point their AI apps to this new service by changing just a single line of code.

We've seen that killer apps for AI combine the power of large language models with your own data to create powerful interactive assistants. This is the key reason why retrieval-augmented generation (RAG) patterns have become the most popular architectural patterns for building AI apps. Last week, OpenAI introduced a new Assistants API at their developer day to handle all the stateful operations required for large language model-based applications, including persistent threads and messages, files, and automatic RAG. 

We think that unifying data and LLMs within a single API is a great idea, so we took a look at how to support this within our Astra DB Vector database. Today we're introducing our Astra DB Assistants API, which is a drop-in replacement for the OpenAI Assistants API. The end result is a simple, scalable datastore for your AI apps that's backed by the power of Cassandra, so you can build LLM-powered assistants with the performance, reliability, and cost advantages of the world's most scalable vector database. We're making it available as a developer preview today so that your API applications can switch to using this API by changing a single line of code!

To see it in action and see for yourself how easy it is to start building with it, watch the video below and read this getting started post.

Try the Astra Assistants API 

To try the Astra Assistants API, create your free Astra account, and start here. 

Make a copy of this OpenAI tutorial notebook, change one line of code, and run through all of the API operations in the notebook against the Astra DB backend.

With huge respect for OpenAI, we're excited to be the first cloud company offering a compatible API.

By enabling this preview, you agree to the DataStax Preview Terms.