CustomizeData.md

April 30, 2026 ยท View on GitHub

Customize the solution with your own data

If you would like to update the solution to leverage your own data please follow the steps below.

Note: you will need to complete the deployment steps here before proceeding.

Prerequisites:

  1. Your data will need to be in JSON or wav format with the file name formated prefixed with "convo" then a GUID followed by a timestamp. For more examples of the data format, please review the sample transcripts and audio data included here

    • Example:
      • Transcripts: convo_32e38683-bbf7-407e-a541-09b37b77921d_2024-12-07 04%3A00%3A00.json
      • Audio: convo_2c703f97-6657-4a15-b8b2-db6b96630b2d_2024-12-06 06_00_00.wav
  2. Navigate to the storage account in the resource group you are using for this solution.

  3. Open the data container

Note for WAF-aligned deployments: If your deployment uses private networking, you'll need to log into a VM within the virtual network to upload files. See VM login instructions below.

  1. If you have audio files, upload them to custom_audiodata folder. If you have call transcript files, upload them to custom_transcripts folder.

  2. Navigate to the terminal and run the process_custom_data.sh script to process the new data into the solution with the following commands:

    bash ./infra/scripts/process_custom_data.sh
    

    If you don't have azd env then you need to pass parameters along with the command. Parameters are grouped by service for clarity. The command will look like the following:

    bash ./infra/scripts/process_custom_data.sh \
      <Resource-Group-Name> <Azure-Subscription-ID> \
      <Storage-Account-Name> <Storage-Container-Name> \
      <SQL-Server-Name> <SQL-Database-Name> <Backend-User-MID-Client-ID> <Backend-User-MID-Display-Name> \
      <AI-Search-Name> <Search-Endpoint> \
      <AI-Foundry-Resource-ID> \
      <OpenAI-Endpoint> <Embedding-Model> <Deployment-Model> \
      <CU-Endpoint> <CU-API-Version> <AI-Agent-Endpoint> <Solution-Name>
    

How to Login to VM Using Azure Bastion

For WAF-aligned deployments with private networking:

  1. Navigate to your VM in the Azure portal
  2. Click Connect โ†’ Bastion
  3. Enter your VM credentials (username and password) and click Connect
  4. Wait for the Bastion connection to establish - this may take a few moments
  5. Once connected, you'll have access to the VM desktop/terminal interface