Fabric Data Agents currently do not allow developers to configure or swap the underlying LLM, which limits flexibility compared to Azure AI Foundry and Copilot Studio.
To better support advanced scenarios, consider:
- Allowing custom model selection
- Supporting hybrid patterns (e.g. bring your own model for NL→SQL generation)
This would:
- Reduce the need to wrap Fabric Data Agents in external orchestration layers
- Enable advanced optimisation for domain-specific use cases
No CommentsBe the first to comment
