Microsoft enables digital transformation for the era of an intelligent cloud and an intelligent edge. Its mission is to empower every person and every organization on the planet to achieve more. Microsoft is dedicated to advancing human and organizational achievement. Microsoft Security helps protect people and data against cyberthreats to give peace of mind.
Do You Manage Peer Insights at Microsoft?
Access Vendor Portal to update and manage your profile.
The strongest point for me, and there are a host of things to explore, is the integration with Azure Databricks for feature store workflow. I have been able to build feature tables and Delta live tables in Databricks, and Foundry does relatively well with navigating them. In my own process, I had to move around from platform to platform a great deal, so I appreciate Foundry's ability to provide one unified place for data integrity and experimentation.
Unified AI resource to manage models, endpoints and deploy applications. Model catalog is great and you have variety for choosing the best model that fits into each AI use case and control costs. Improving constantly the agents deployments and integration with OpenAI libraries when deploying a gateway.
End-to-end AI application workflows covering experimentation, evaluation, development and monitoring in a single platform. Strong integration with Azure services, making it easier to manage security, access control and scalability. Built-in governance, guardrails and observability features that make enterprise AI development safer and more reliable.
I'm not directly involved in these conversations, but it has been impressed upon me that cost management is imperative. We often have to consult with our tenant managers to ensure that spinning up resources in Databricks and Foundry is tracked with disciplined tagging or it can result in unexpected costs. It has certainly made me more diligent in documentation and the process of our internal workflow.
Stability of the resource, Microsoft recently launched a new foundry with tons of bugs but with new features that legacy foundry was not offering. Since many features are very recent we are lacking the terraform provider implementation of some resources. Forces you to have App Insights (with high cost bill) to have a complete observability of the resource.
Initial learning curve can be challenging for teams without prior Azure experience. Cost tracking and usage transparency could be more intuitive during large scale experimentation. Some advanced configurations require deeper documentation or more guided templates.