Problem
Nerdio Manager Copilot has historically been tied to a specific Azure OpenAI model (for example gpt-35-turbo / version 0125). When that model was deprecated in Azure, Copilot enablement in our environment failed with ServiceModelDeprecated errors until we upgraded to Nerdio Manager 7.5.1, which removed GPT-3.5. This creates two issues:
- Tight coupling to a single model/version – any future deprecation (or capacity/region constraints) can break Copilot with no customer-side mitigation.
-
No way to choose the best model for our use case – we can’t take advantage of newer models like OpenAI’s
gpt-5.1/gpt-5.1-chat, or alternative Azure AI Foundry models like Anthropic’sclaude-sonnet-4-5, even when they are available and preferred for enterprise assistant scenarios.
As one example, after successfully enabling Copilot on 7.5.1, the prompt:
“What’s the best practice recommendation for managing FSLogix configuration and RDP configuration across multiple workspaces and host pools?”
returned the generic response “I couldn’t find answer for your question. Please, try to ask it differently.” For an AVD-focused product, we’d like the option to move Copilot onto newer, stronger models as they become available.
Description
Please decouple Nerdio Manager Copilot from a single hard-coded Azure OpenAI model and instead allow admins to configure which Azure AI Foundry model Copilot uses, at least at the account level (and ideally per tenant / per workspace).
Concretely, in the Copilot configuration UI and/or via Nerdio Manager settings, expose:
- Model provider / family (e.g., “Azure OpenAI – OpenAI models”, “Azure AI Foundry – Anthropic models”).
-
Model ID string, matching the Azure AI model catalog (for example:
-
gpt-5.1orgpt-5.1-chatfor the latest OpenAI models in Azure AI Foundry, -
claude-sonnet-4-5for Anthropic Claude Sonnet 4.5, - or future models like
gpt-5-pro, etc.). Azure AI+2Azure AI+2
-
- Optional deployment name / endpoint where relevant, so customers can point Copilot at an existing enterprise deployment rather than having Nerdio create a specific one.
This would keep Copilot aligned with the current Azure catalog (including the newer GPT-5.1 generation and partner models) without waiting for a Nerdio release whenever Microsoft retires a model or introduces a better one. TECHCOMMUNITY.MICROSOFT.COM+2OpenAI Platform+2
Vision
From an admin point of view:
- In Settings → Copilot (or equivalent), provide a simple configuration section:
- “Model Provider” (dropdown):
- Azure OpenAI – OpenAI
- Azure AI Foundry – Anthropic
- (future providers as added)
- “Model ID” (text box with auto-complete from the Azure AI catalog), e.g.:
-
gpt-5.1-chat(preferred default going forward), or -
gpt-5.1for heavier reasoning, or -
claude-sonnet-4-5for some customers that standardize on Anthropic.
-
- “Deployment name / endpoint” (optional override) for customers who want to re-use an existing deployment instead of letting Nerdio create/manage it.
- “Model Provider” (dropdown):
- Nerdio Manager Copilot uses this configuration whenever it:
- Creates or updates the backing Azure deployment.
- Sends chat/completion requests on behalf of the admin.
- When Microsoft retires a model (for example, if
gpt-5.1-chatis ever replaced by a newergpt-5.2-chat), admins can:- Create the new deployment in Azure,
- Update the model ID / deployment reference in Nerdio Manager,
- And keep Copilot working without waiting for a new Nerdio release.
- Optionally, expose the currently used model and version in the Copilot UI footer or “About” panel so support and admins can quickly verify which model Copilot is running on.
This change would:
- Make Copilot resilient to Azure model deprecations and region changes.
- Let customers adopt newer models like
gpt-5.1/gpt-5.1-chatas they become generally available. - Enable advanced users to standardize on other enterprise models from Azure AI Foundry such as
claude-sonnet-4-5while still benefiting from Nerdio’s Copilot UX.
Comments (2 comments)