How to update an index for a data source in Azure AI Studio?

Copper Contributor


I have a functioning deployment and web app for a RAG application that I created in Azure AI Studio.  Now I want to update the app to use additional data I have added.  I figured out how to add another version of my data in Azure AI Studio, but can't figure out how to add another version of my index. It seems like versioning is supported. The documentation I found helps with creating data sources/indexes, but not updating.


Help answering this or pointing in another direction would be great.


6 Replies

@Peter McNally 

I have the same exact issue and needed to put in a support request today. Available documentation shows ways to update an index using API requests but that does not suffice.


Using the same AI search index in the playground will overwrite the existing index. I will get back to you once I get a fix.

@lumitekinc Ever figure this one out?

Yes I did. I upload new data to my data store (storage account).

Go to the AI search, look for the indexer and run the index manually.
You can also create a schedule for the index. Azure AI Studio is just a GUI for many services, one of which is a job in Azure ML Studio that creates the index. You can schedule the job there. I am adding additional data sources to my data store as needed. I have a job in ML Studio to update the index each night.
Thanks Peter - this worked a treat. Creates a new version of the index, chunks up the new PDF's from the source data. Exactly what I wanted. This is probably enough to get a v1 shipped, thanks for the steer.
I'm assuming that this doesnt chunk up the source data into chunks so the LLM can work within its context window? Let me know if it's a bad assumption. Re running the ML job works, but I'm not sure how to set up the indexer for the existing index (created in Azure AI Studio) in Azure AI Search