AI solutions built to power industrial innovation and sovereign control | OD839
Inbal Sagiv explains how Azure AI Foundry can be brought to industrial and sovereign environments using Foundry Local on Azure Local, with Azure Arc providing a consistent management and governance layer. The session focuses on running AI apps and agents with low latency and local data control, including connected and fully disconnected scenarios.
Overview
This Microsoft Build 2026 session introduces Foundry Local on Azure Local as an approach for building and running AI applications directly on local infrastructure, including environments that require:
- Low-latency inference close to machines and operations
- Local data control for sovereignty and regulatory constraints
- Resilience for sites with unreliable connectivity
- Fully disconnected operations where cloud access is not possible
A key theme is maintaining a consistent developer and governance experience by using Azure Arc to manage and govern these local deployments.
What the session covers
From applications to agentic AI
- The session frames an industry shift toward agentic AI (AI agents that can use tools and perform multi-step tasks).
- It connects this shift to industrial and sovereign scenarios where reliability, control, and locality matter.
Sovereign AI and sovereign environments
- The session discusses Microsoft’s positioning around Sovereign AI and sovereign cloud concepts.
- The focus is on enabling AI workloads in environments with strict requirements around data residency, control, and operational independence.
Azure Local as the on-premises foundation
- Azure Local is presented as the infrastructure foundation for running AI locally.
- The session positions Foundry Local as a way to run Foundry capabilities on top of Azure Local for both connected and disconnected use cases.
Foundry Local capabilities highlighted
The session calls out several capabilities for Foundry Local, including:
- Multi-node support
- Local RAG (retrieval-augmented generation with local data)
- Custom tool integration (for agent/tool workflows)
Models and runtimes
- Overview of a model catalog and a bring-your-own-model (BYOM) approach.
- Discussion of inference runtimes used to run models locally.
Demo: deploying and managing models locally
- A technical demo shows deploying and managing models using:
- A command-line workflow
- The Azure Local interface
Building local AI agents
- The session covers developing local AI agents, including agentic RAG patterns.
- Demos mentioned include chat and video agent scenarios.
- The session includes preview registration information for Foundry Local on Azure Local.
Resources
- Session page: https://aks.ms/build26/OD839
- Foundry Discord: https://aka.ms/build/foundrydiscord
- Build 2026 blog hub: https://aka.ms/build26blog
- Docs: https://learn.microsoft.com/en-us/azure/azure-sovereign-clouds/private/foundry-local/what-is-foundry-local-on-azure-local
- Preview request: https://aka.ms/FoundryLocalAzure_PreviewRequest