At the same time, many AI services have recently been launched or rebranded. Some readers may be wondering how Microsoft Foundry differs from other Microsoft services, such as Microsoft 365 Copilot and Copilot Studio.
This article provides a clear explanation of Microsoft Foundry, including its role, key features, practical use cases, and how to get started.
Contents
1. What Is Microsoft Foundry?
Microsoft Foundry is a unified platform that brings together the capabilities needed to develop and operate AI applications and AI agents. Let us begin with an overview of the platform and the background behind its name change.
1-1. The Concept of an “AI App and Agent Factory”
Microsoft Foundry is positioned as an “AI app and agent factory.” Its defining feature is that model selection, agent development, connection to internal company data, and post-deployment monitoring and governance can all be managed on a single platform.
With Microsoft Foundry, companies can select the right models from a wide range of AI options, build proprietary AI applications and agents connected to internal data and existing systems, and move them into enterprise-wide production while maintaining security and governance.
1-2. The Name Change from Azure AI Foundry and Its Background
Microsoft’s AI development platform debuted as “Azure AI Studio” in 2023. It became “Azure AI Foundry” in November 2024, and the current name was announced at Microsoft Ignite in November 2025 before becoming official in January 2026. “Azure” was removed from the name because the offering had evolved beyond the cloud into a broader Microsoft AI platform that also supports AI execution on PCs and edge devices.
Along with the name change to Microsoft Foundry, the service expanded beyond its previous emphasis on OpenAI models. It now supports many models, including third-party options such as Claude, while also strengthening the governance capabilities required for enterprise use.
1-3. How It Differs from Microsoft 365 Copilot
Microsoft 365 Copilot is a ready-to-use AI assistant, while Microsoft Foundry is a platform for building AI tailored to your organization. Copilot may be sufficient for improving productivity in Word and Excel. When a company needs AI to align with its own operations and systems, Microsoft Foundry becomes the more appropriate choice.
It is also important to understand how Microsoft Foundry differs from Copilot Studio, which enables low-code AI development. In Copilot Studio, users can build internal chatbots, customer support solutions, and similar tools through a visual interface. Copilot Studio is well suited to business teams that want to build solutions quickly, while Microsoft Foundry is better suited to developers who need deeper control over model selection, external integrations, and solution design.
2. Key Components of Microsoft Foundry
Microsoft Foundry consists of multiple components with distinct roles. The six main components are summarized below.
| Component | Role |
|---|---|
| Foundry Models | A catalog for selecting and using AI models from multiple providers |
| Foundry Agent Service | A platform for building and running AI agents |
| Foundry IQ | Connects internal company data as knowledge through RAG |
| Foundry Tools | A collection of tools for web search and external integrations |
| Foundry Local | Runs AI locally on PCs and edge devices |
| Foundry Control Plane | Provides visibility into and governance over AI use across the organization |
2-1. Foundry Models (Model Catalog)
Foundry Models is a catalog of more than 11,000 AI models as of July 2026. It includes OpenAI’s GPT family, Anthropic’s Claude family, Meta’s Llama, and Microsoft-developed models such as MAI and Phi. Organizations can compare performance and select models based on their use cases and budgets, enabling architectures that are not dependent on a single vendor.
2-2. Foundry Agent Service (Agent Runtime Platform)
Foundry Agent Service is a managed platform for building and running AI agents. Users can choose an implementation approach, including “prompt agents,” which can be created through on-screen configuration, and “hosted agents,” which support custom code. Because it is managed, there is no need to maintain servers. Identity management, activity logging, and monitoring capabilities are built in, allowing the same environment to be used for production operations.
2-3. Foundry IQ (Using Internal Company Data)
Generative AI does not automatically understand a company’s policies or manuals. Foundry IQ allows organizations to register internal data in locations such as SharePoint as a knowledge base that agents can reference when answering questions. This enables RAG, or retrieval-augmented generation, without requiring organizations to build it from scratch. Agents can then use internal company data while respecting access permissions.
2-4. Foundry Tools (External Service Integration)
Foundry Tools gives AI agents the capabilities needed to take action. In addition to built-in tools such as web search and code execution, it supports Model Context Protocol (MCP), a standard for external integration. This allows agents to connect with internal systems and SaaS applications and carry out business processes.
2-5. Foundry Local (Local AI)
Foundry Local enables AI models to run on PCs and edge devices without going through the cloud. It is expected to be useful for operations where confidential data cannot leave the environment and for work performed in offline settings.
2-6. Foundry Control Plane (Operations and Governance Platform)
Foundry Control Plane provides centralized visibility into the use and cost of agents and models across an organization and enables policy enforcement. It supports AI guardrails, security, and operational monitoring of agents.
3. Business Use Cases Enabled by Microsoft Foundry
Microsoft Foundry makes it possible to apply AI closely aligned with a company’s operations, including scenarios that are difficult to address with off-the-shelf AI products. Four representative examples are described below.
Event-Driven, Autonomous Back-Office Agents
An incoming invoice email can trigger an agent to read the invoice, create the necessary record, and notify the person responsible, all without human intervention. This is a persistent AI agent that can operate without waiting for user instruction.
Embedding AI Capabilities into Your Own Products and Services
Organizations can expose an agent through an API and embed it in their own services or customer portals. This makes it possible to offer capabilities such as automated quotation creation and support bots directly to customers.
Offline AI in Closed Environments, Such as Factories
With Foundry Local, manufacturing sites with restricted networks can search equipment manuals and summarize daily reports directly on local devices. This approach is well suited to operations where confidential data cannot be sent outside the environment.
A Proprietary Model Tuned with Your Company’s Data
A model can be adjusted using data such as past quotations and support histories, creating AI that understands industry-specific terminology and the company’s own rules.
4. How to Get Started with Microsoft Foundry
If you have an Azure account, you can try Microsoft Foundry using only a web browser. The basic process is as follows:
- Prepare an Azure account.
- Access the Foundry Portal at ai.azure.com
- Create a project and test a model.
- Build an AI agent, then connect to internal data and tools.
The platform itself is free to use. Pricing is generally consumption-based and depends on factors such as the AI model usage volume, measured in tokens, and related search services. This allows organizations to begin with a small proof of concept and expand in phases.
Model pricing and regional availability are updated frequently. For the latest information, please refer to Microsoft’s official website or contact SYSCOM.
Conclusion
Microsoft Foundry is an integrated platform for building and operating AI applications and AI agents. It succeeds and expands the former Azure AI Foundry, bringing together a broad selection of AI models, an agent platform, connections to internal company data, and governance capabilities in one environment.
At the same time, deciding which model to use, what can be handled with Copilot, and where deeper development in Microsoft Foundry is needed requires an understanding of the broader Microsoft product ecosystem.
If your organization is considering Microsoft Foundry or other approaches to AI adoption, please contact SYSCOM GLOBAL SOLUTIONS. For more than 30 years, we have supported our customers as a trusted Microsoft Solutions Partner, helping them navigate and address their business challenges.
We help our customers improve operational efficiency by recommending the right AI and digital transformation solutions for their specific industry and business needs. Contact SYSCOM to discuss how we can help.