How to backup Azure AI Assistant and Chat in Azure Foundry (aka AI Studio) Playground?

Question

Hi,

Our team is using Azure AI Assistants and Chat via the Azure AI Foundry Playgrounds. Lot of work is being done with system prompts, parameters, data being uploaded (search and blobs storage), even prompt engineering by the business units. Hence, we have to treat Azure AI Foundry playgrounds as production.

We will be using only managed services for now (no container based deployment for now).

We want to be able to back up the Azure AI services, data being uploaded, configurations, etc.

Q) How do we do that backup Azure AI Assistant and Chat?

Chat playground has minimal export to a JSON file for system prompts, parameters and few shot prompts. However, import is flaky. Assistant (in preview) currently has no exports. As business unit users could be uploading different files, change prompts, I need a decent way to back up and recover (RTO, RPO and MDT being tracked).

Answer 1

Hi there Gaining Knowledge

the Chat Playground allows limited JSON exports for system prompts, you can supplement this by creating scripts using Azure SDKs to programmatically back up prompts and configurations into a version-controlled reposs. For uploaded files, make sure all data is in Blob, and use Azure Backup o for regular backups. If your Assistants integrate with databases like Azure SQLcosmos, schedule backups using their built-in features. Enable Azure Monitor and Application Insights to log changes and maintain version tracking.

If this helps kindly accept the answer thanks much.

Answer 2

Hello Gaining Knowledge,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that you would like to know how you can backup Azure AI Assistant and Chat in Azure Foundry (aka AI Studio) Playground.

For a robust backup and recovery strategy for the Azure AI Foundry Playground, you can follow the below steps:

For Chat Playground, automate the export process to a secure storage location by using the Playground's export feature via API (if available) or scripts to periodically save configurations (e.g., system prompts, few-shot examples) as JSON files. Address flaky imports by validating exported files and reformatting them if necessary to ensure compatibility. - https://learn.microsoft.com/en-us/azure/ai-studio So, for Assistant Playground, as no export feature is available, raise a support ticket with Azure to request access to APIs for exporting configurations or consider manual documentation of key configurations until APIs are available.

Secondly, leverage Azure Blob Storage for user-uploaded files:

• Enable versioning and soft delete to recover from accidental overwrites or deletions.
• Schedule regular Azure Backup jobs for the storage account to ensure recovery within RTO and RPO objectives.

https://learn.microsoft.com/en-us/azure/storage/blobs/versioning-overview and https://learn.microsoft.com/en-us/azure/storage/blobs/soft-delete-overview

Thirdly, use Azure Resource Manager (ARM) Templates or Bicep to capture the configurations of all Azure resources associated with the Playground, such as AI services, storage accounts, and networking setups. For frequent configuration changes, integrate Azure DevOps or GitHub Actions pipelines to export configurations regularly and store them in a secure repository. - https://learn.microsoft.com/en-us/azure/azure-resource-manager/templates/overview and https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep

Also, set up Azure Monitor to track changes to key resources like Blob Storage, AI configurations, and permissions. Create alerts for anomalies or unauthorized changes. Use Change Tracking and Inventory (part of Azure Automation) to monitor changes in configurations or uploaded files.

To implement RTO, RPO, and MTD Testing:

• Conduct monthly recovery drills by simulating a disaster and verifying the recovery process for data and configurations.
• Create a playbook for backup and recovery, detailing each step, tools used, and stakeholders responsible.

https://learn.microsoft.com/en-us/azure/site-recovery/site-recovery-overview

Finally, these are supplementary tools and best practices solution:

• Use Azure Data Factory to periodically copy data from Blob Storage to a secondary location for redundancy. - https://learn.microsoft.com/en-us/azure/data-factory/introduction and https://learn.microsoft.com/en-us/azure/data-factory/copy-data-tool-introduction
• Investigate third-party backup tools, such as Cohesity or Veeam, which may offer better granular control over Azure resource backups, including file-level and configuration-lev backups. - https://www.cohesity.com/solutions/microsoft and https://www.veeam.com/microsoft-azure-backup.html

I hope this is helpful! Do not hesitate to let me know if you have any other questions.

---

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.