Summary
At their 2025 Pre-Briefing, Microsoft laid out a clear vision for how Agentic AI is becoming the next foundational layer of its platform strategy. Instead of treating agents as optional enhancements, the company is positioning them as central components that connect models, data, developer tools, and governance into one system. Yina Arenas and Amanda Silver walked through how Microsoft Foundry, GitHub, VS Code, and the broader Azure and Microsoft 365 ecosystem fit into this shift.
Their core message was consistent throughout: enterprises are moving beyond simple chat-based interactions and now require agents that can carry out tasks, coordinate actions, and follow policies across different environments. Microsoft aims to provide a unified platform where these agents can be built, deployed, and managed at scale.
Foundry Reframed as Microsoft’s Agent Platform
A central update was the rebranding of Azure AI Foundry to Microsoft Foundry, reflecting a tighter alignment with Microsoft 365 and Microsoft Fabric. Microsoft wants customers to view these three areas as a connected system, where Microsoft 365 is the workplace interface, Fabric is the governed data foundation, and Foundry is the platform that powers agents and AI applications.
Foundry is already widely used, both inside and outside Microsoft. Tens of thousands of organizations rely on it today, and nearly all of Microsoft’s internal AI-powered products run on top of it. That breadth of usage gives Microsoft confidence in positioning Foundry as an enterprise-ready platform for teams moving beyond experimental projects.
At its core, Foundry combines a large model catalog behind a single API, a runtime for deploying and orchestrating agents, a retrieval layer that helps agents access enterprise knowledge, and a control plane that provides security, observability, and lifecycle management. Microsoft’s aim is to remove the need for customers to stitch together separate tools just to get agents into production.
Renaming the platform to Microsoft Foundry acts as more than just a brand update. Microsoft is signaling that Foundry is intended to become a foundational layer, similar to how Azure became central to cloud adoption years ago. This time, the focus is on agents rather than infrastructure.
Models, Fine-Tuning, and Local Deployment
The model strategy within Foundry continues to emphasize flexibility. Microsoft now offers access to an extremely broad set of models through a single API. With the Model Router moving to general availability, teams can rely on the system to direct requests to the best available model without rewriting code.
Fine-tuning options now support multiple training methods and model providers, and synthetic data capabilities continue to grow. The expansion of Foundry Local, which includes support for Windows, macOS, and Android, allows these models to run offline. This is increasingly important for industries that need AI at the edge or in restricted environments.
Many enterprises worry about choosing the “wrong” model or losing flexibility over time. Foundry’s approach, combining a unified API, the model router, fine-tuning, and local runtime, aims to reduce that risk and give teams a future-proof path.
Giving Agents Context, Action, and Control
Microsoft spent significant time on what enterprises struggle with most once they begin adopting Agentic AI: giving agents the right information, enabling them to take action, and keeping them governed.
Foundry IQ provides the context layer, connecting agents to SharePoint, OneLake, Fabric, databases, internal applications, and the open web. Instead of building custom retrieval pipelines, teams can rely on Foundry IQ to manage query planning, data access, and security through Microsoft Purview and Entra. This reduces both complexity and risk.
Foundry Tools lets agents perform meaningful actions. With a catalog of AI skills and connectors across common business systems, agents can update records, trigger workflows, generate content, or interact with internal and external applications. Organizations can also add their own APIs as custom tools, making Foundry an extension of their existing systems.
The Foundry Control Plane is the operational centerpiece. It provides visibility into what agents are doing, how they are performing, and whether they are following policy. It supports identity and access management with Entra Agent ID, built-in threat detection, data protection through Purview, and full lifecycle controls for starting, stopping, updating, or quarantining agents.
Building an agent is not the hard part, however, running many agents responsibly is. Microsoft’s focus on context, action, and governance shows a clear understanding of what enterprises actually need to move Agentic AI into production safely.
How Developers Will Build and Operate Agents
Microsoft outlined plans to make Agentic AI part of the everyday developer workflow through GitHub and Visual Studio Code. The AI Toolkit for VS Code now allows developers to create and test agents directly against the Foundry Agent Service. GitHub Copilot can generate the initial structure for agents using Microsoft’s recommended patterns. A new plan mode helps teams agree on the architecture of an application before producing code, improving consistency and reducing rework.
Agent-driven code review, based on Microsoft’s internal practices, brings automated and context-aware review to pull requests. GitHub is also becoming the main hub for tracking agent tasks, viewing activity, and managing guardrails.
Modernization was another major theme. Microsoft shared internal examples where teams used Copilot and Foundry-aligned tools to upgrade large sets of Java and .NET applications much faster than traditional methods. These upgrades tie directly into Azure App Service and Azure Migrate, creating a cleaner path from legacy systems to Agentic AI-enabled environments.
Developer experience will be a major differentiator among AI platforms. GitHub and VS Code are deeply embedded in enterprise development, and bringing Agentic AI directly into those workflows gives Microsoft a strong advantage.
Looking Ahead to Ignite
Microsoft wants Agentic AI to become a core layer of enterprise architecture, not a feature scattered across products. Foundry sits at the center, supported by Microsoft 365, Fabric, GitHub, and Azure.
As Ignite continues, we’ll be watching for real customer stories that show how these capabilities work together in practice. Key questions include how smoothly Foundry integrates with existing development processes, how well the control plane handles mixed environments, and how reliably agents perform when deployed broadly.
Microsoft is betting that enterprises will succeed with AI only when agents become manageable, trustworthy, and tightly connected to the systems employees already use. This briefing suggests that Foundry is the platform Microsoft intends to use to make that future possible.
Twilio Report Exposes Trust Gaps in Enterprise Conversational AI
New data highlights rapid adoption of conversational AI in addition to a widening divide between…
Google Unveils Nano Banana Pro Advancing Image Generation
Google Public Sector Summit Signals a New Phase of AI-Driven Government Modernization
Twilio Posts Record Q3 as Customer Growth and Profitability Strengthen
AMD and DOE Forge the Foundation for America’s Sovereign AI Supercomputing Future
