If you have seen Azure AI Studio in a tutorial, you may be wondering whether that name still points to the tool Microsoft offers today. Microsoft’s AI product names have changed, which can make older lessons hard to follow. The platform now supports work with AI models, applications, and agents, with options for both new and experienced developers.
This guide helps you decide how to get started. You will see what the current platform can do, how it fits with other Microsoft products, and which first steps suit your experience. Customers can explore ideas without having to begin with a large project or a complex setup.
Key Takeaways
- Microsoft Foundry is the current platform name.
- Azure AI Studio is a former name, not a separate cloud.
- Foundry supports model exploration, AI applications, and agents.
- A visual path and a code-first path are both available.
- Copilot Studio suits managed low-code work; Foundry offers developers deeper control.
- Start with one clear task and test it with representative data.
Treat the older name as a useful clue to the product’s history, not as the name to look for in a current project.
What Is Azure AI Studio Called Today?
The name to use for the current platform is Microsoft Foundry. It is the AI development platform within Microsoft Azure, the broader cloud service that also offers computing, storage, databases, and many other Azure services.
Azure AI Studio was a development experience within Microsoft Azure, not a competing cloud platform. The older name can still appear in tutorials, saved projects, or search results, but it does not name a separate cloud from Microsoft Azure.
Use Microsoft Foundry for the current platform name. Azure AI Studio and Azure AI Foundry are former names.
How the platform name changed
Microsoft renamed Azure AI Studio to Azure AI Foundry at Microsoft Ignite in November 2024. At Microsoft Ignite in November 2025, it renamed Azure AI Foundry to Microsoft Foundry.
That sequence explains why a tutorial may say Azure AI Studio, while a newer product page says Microsoft Foundry. In 2026, Microsoft Foundry is the platform for building AI agents and applications with models and tools.
Azure Foundry is older or informal wording, not the current product name. The phrase “Microsoft Azure Studio” can also cause confusion, but it is not the name to use for the current AI platform.
How it relates to Microsoft Azure
Microsoft Azure is the cloud environment. Microsoft Foundry is one platform within that environment, bringing together tools for AI development and deployment.
Think of Azure as the larger set of cloud services and Foundry as an AI-focused work area inside it. That distinction helps when a tutorial mentions both the cloud and an AI tool.
Azure OpenAI Service is now part of Foundry as Foundry Models, which Azure sells. This connects the former service name to the current model offering without making Azure OpenAI a separate cloud.
When following older material, check whether its Azure AI Studio instructions still apply to the current Foundry experience.
What Does Microsoft Foundry Include?
Microsoft Foundry brings several AI-building services into one platform instead of asking a beginner to assemble a stack from unrelated tools. Each component helps with a different part of AI development, from model access to custom machine learning.
In 2026, Foundry provides access to more than 10,000 models from Microsoft, OpenAI, Anthropic, Meta, and other providers. Its model catalog adds about 50 new models each month, but a large catalog does not mean every model suits every task.
The services that make up the platform
Foundry Models provides model selection and evaluation. Foundry Agent Service offers a managed runtime for building agents, while Foundry Tools adds functions such as language and content processing.
Foundry IQ helps connect applications and agents to relevant knowledge. Azure Machine Learning supports custom machine-learning work, such as training and managing models built for a specific need.
| Component | What it helps with | Beginner example |
|---|---|---|
| Foundry Models | Accessing and evaluating models | Compare models for a summary task |
| Foundry Agent Service | Building and running agents | Create an agent that uses a tool |
| Foundry Tools | Adding ready-made AI functions | Analyze text for key details |
| Foundry IQ | Connecting agents to knowledge | Find useful information in team content |
| Azure Machine Learning | Custom machine-learning projects | Train a model for a focused task |
Finding models for different tasks
Foundry Control Plane helps teams manage and govern AI resources across the platform. Foundry Local supports running models locally on a device, which can suit some development or offline needs.
The catalog’s variety gives developers room to compare models, but model names alone do not show how well they handle a specific job. A language model built for chat may not be the right choice for image tasks or specialized business needs.
Separate AI products and services can still fit different teams. A developer may use Foundry for a new application while another group continues to use other Azure services for existing work.
Start with the component that solves the immediate job rather than trying to use every service in the platform.
How Do You Choose a Model and Get Started?
A useful first experiment is to compare two candidate models on the same task and the same representative inputs. Real examples can reveal differences that a model’s name or short description will not show.
Foundry Models lets you compare and evaluate models side by side using real-world tasks and your own data. That can help developers judge quality and fit, but an evaluation alone does not prove a model is ready for production.
Explore models before committing
Choose a task that resembles the work your application must do. For example, ask each model to summarize the same set of support messages, then compare accuracy and usefulness.
Your team’s own data can expose gaps that a generic test misses. Keep private or sensitive information within the rules that apply to your organization.
Model selection is only one part of AI development. You can also read a comparison of standalone AI assistants, but those assistants are not the same as models in Foundry’s catalog.
Choose a visual or code-first workflow
A visual workflow lets you explore models and settings through a user interface. It can be a comfortable way to learn before writing application code.
A code-first workflow gives developers more direct control. Azure AI Studio supported SDKs and Visual Studio Code extensions for local development, and Foundry provides paths for working with code as well.
Visual Studio Code can help you move from a small test to a working application. Developers comparing editor tools can also review AI tools for coding workflows and test them on a defined task.
- Define one task and a clear measure of success.
- Shortlist models that appear suited to that task.
- Test each model with representative inputs.
- Record the result and choose the next experiment.
Evaluation can reveal tradeoffs in output quality and task fit. It does not check every part of deployment, such as how an application handles errors or changing user needs.
Keep the first test narrow enough that a beginner can explain why one result is more useful than another.
What Can You Build With Microsoft Foundry?
A team can use Foundry to build anything from a grounded question-answering app to an AI feature inside an existing application. Retrieval-augmented generation (RAG) retrieves relevant information from approved data and supplies it to a model before the model responds.
RAG can help an application answer from selected business information instead of relying only on general model knowledge. For example, a chatbot can search approved company files before it drafts a reply.
- A knowledge-grounded question-answering app for approved business information.
- A customer-service chatbot that helps answer common support questions.
- An internal document search assistant for employees.
- A language or content-processing application that reviews text.
Knowledge sources might include Microsoft SharePoint or OneLake in Microsoft Fabric. Other connections can include Microsoft Bing, Azure Cosmos DB, and Azure Functions.
An application could use Azure Functions to take an action, such as calling a business process after a user asks for help. Microsoft’s RAG chat app sample shows a document-based chat design with answer citations and multi-turn chat.
The right source depends on what the app needs to answer and what information its users may access. A customer-facing chatbot should not draw from the same unrestricted material as an internal assistant.
Match the data source to the information the application is allowed and expected to use.
How Does Foundry Agent Service Work?
A prompt agent is the lighter starting point when you want Foundry to manage the runtime while you configure its instructions, model, and tools. Foundry Agent Service also supports voice-based prompt agents and hosted agents.
Prompt agents need configuration, not application code or containers. Foundry manages their runtime, so you can focus on the agent’s purpose and the tools it may use.
Prompt agents and hosted agents
Hosted agents run customer-provided code or frameworks inside containers. Foundry supplies a managed endpoint, automatic scaling, identity, and observability for that hosted work.
This changes who manages the runtime. With a prompt agent, Foundry handles it; with a hosted agent, your code or framework runs in the managed container environment.
Hosted agents support Agent Framework, LangGraph, the OpenAI Agents SDK, and the Anthropic Agent SDK. Those options can help a team use a framework that fits its existing development approach.
Frameworks, tools, and publishing options
Agent toolboxes can include web search, file search, and code interpreter. They can also include MCP servers, which connect agents to tools through the Model Context Protocol, or custom functions built for a specific task.
Giving an agent tools changes what it can do, so choose tools that fit its job. A web search tool may help answer current questions, while a custom function can connect to a team’s own service.
Agents can be published for Microsoft Teams, Microsoft Copilot, and the Entra Agent Registry. The place you publish an agent affects how customers or employees reach it.
Agent features differ from coding assistants that suggest changes as you work. For a broader view of developer assistants and coding agents, compare their tasks and how much work they can take on.
A prompt agent can be a good fit when configured instructions and tools cover the job. Choose a hosted agent when the customer-provided code or framework is essential to how the agent must run.
How Does Microsoft Foundry Handle Security and Monitoring?
Foundry’s governance story starts by bringing models, agents, and tools under a single Azure resource with shared access controls and policies. Shared controls help teams manage access and networking in one place.
Foundry Agent Service supports Microsoft Entra identity, role-based access control, content filters, and virtual network isolation. These safeguards help control who can use resources and how an agent handles content and network access.
Access control and monitoring solve different problems. Access rules set who can do what, while monitoring helps teams see how an agent behaves during use.
For visibility, Foundry Agent Service supports end-to-end tracing, metrics, evaluations, and integration with Application Insights. Tracing can help a team follow an agent’s steps, while metrics show patterns in its activity.
Enterprise teams can use evaluations to check how outputs perform against chosen measures. Tracing and monitoring can help identify issues that a single test did not reveal.
Some enterprise-readiness capabilities are in preview. A preview feature may change or may not be available in every setup, so a team should not make it a production dependency without checking its status.
Security settings still need to match the data and users involved. An internal assistant with access to company files needs rules that fit those files and the people who can use them.
Confirm which monitoring and evaluation capabilities are generally available before setting a production requirement.
Microsoft Foundry vs Copilot Studio: Which Fits Your Project?
Choose Copilot Studio when a managed low-code experience and Microsoft 365 integration matter most; choose Foundry when developers need deeper control of the AI stack. The project’s owner and the level of customization help point to the better fit.
Copilot Studio suits managed low-code work, while Microsoft Foundry suits projects that need more control over models, orchestration, and deployment. Azure AI Studio vs Copilot Studio is a comparison between Foundry’s former name and a separate low-code product.
| Option | Best fit | Main tradeoff |
|---|---|---|
| Copilot Studio | Managed low-code apps with Microsoft 365 | Less control over the full AI stack |
| Microsoft Foundry | Developer-led AI applications and agents | More development work to manage |
| Copilot Studio with Foundry behind it | User experience plus advanced AI services | Teams must coordinate the two platforms |
Copilot Studio includes connectors, governance, and Power Platform capabilities. It can be the more approachable choice when a team needs a user-facing copilot without building every part of the system.
Foundry fits developers who need custom models, advanced retrieval-augmented generation, or CI/CD, a way to automate software testing and deployment. It also gives developers more control over orchestration, the steps that coordinate models and tools.
Azure AI Foundry vs Copilot Studio follows the same comparison: Azure AI Foundry is an older name for Microsoft Foundry, not another current platform. The two products can work together, with Copilot Studio providing a user-facing experience and Foundry supplying advanced AI capabilities behind it.
Decide which team owns the user experience and which team needs control of the underlying AI capabilities.
How Does Azure Compare With AWS for AI?
In Q1 2026, AWS held 28% of the global cloud infrastructure market and Microsoft Azure held 21%, but that comparison alone cannot choose a platform for your project. Market share describes cloud use, not whether a service fits your application.
Neither cloud is automatically better for every AI project. The fit depends on existing cloud commitments, the models and tools you need, and the development environment around the work.
Azure AI Studio is an earlier name in the Azure ecosystem. AWS offers Amazon Bedrock for managed access to foundation models from Anthropic, Meta, Mistral, and other providers.
| Comparison point | AWS | Microsoft Azure |
|---|---|---|
| AI model service | Amazon Bedrock | Microsoft Foundry, including Foundry Models |
| Q1 2026 global cloud infrastructure market share | 28% | 21% |
| Q1 2026 overall market context | $129 billion total market; 35% year-over-year growth | $129 billion total market; 35% year-over-year growth |
Amazon Web Services launched in 2006. Its long history does not settle which cloud will work better for your application or your team.
The Q1 2026 market shares show that AWS led Azure in that snapshot. They do not predict whether Azure will overtake AWS later.
Organizations do not need to be leaving AWS as a group for another cloud to suit a particular project. Cloud choices are workload- and ecosystem-specific, and teams may use different providers for different needs.
For a wider look at AWS, Azure, and Google Cloud, compare services, costs, team skills, and existing systems. Market share alone cannot show the full fit.
Compare the services against your actual workload and existing cloud environment, not market share alone.
Can You Still Use Azure Data Studio?
Azure Data Studio retired on February 28, 2026, so it is no longer a maintained tool. Visual Studio Code with the MSSQL extension is Microsoft’s supported replacement.
Azure Data Studio no longer receives updates, security fixes, or maintenance. You can move existing queries, scripts, and SQL database projects without converting them.
Visual Studio Code with the MSSQL extension supports SQL Server, Azure SQL Database, Azure SQL Managed Instance, and SQL database in Fabric. These targets cover common SQL work across local and cloud databases.
SQL Server Management Studio remains supported for SQL Server Agent and general SQL Server administration. Keep it for those tasks while moving other daily work to the replacement editor.
Move daily database work to a supported editor before relying on the retired application.
Use this short checklist to make the change:
- Install Visual Studio Code.
- Add the MSSQL extension.
- Reconnect to the relevant database.
- Keep SQL Server Management Studio for SQL Server Agent and general administration.
Try a few everyday queries and open a SQL database project in the new editor. A small test can show whether your usual workflow is ready to move.
Before uninstalling the old application, confirm that your everyday queries and database projects open as expected in Visual Studio Code.
Conclusion
The practical takeaway is that Azure AI Studio is the former name for Microsoft Foundry, the current platform for building AI applications and agents on Azure. Your starting point depends on whether you need a simple low-code experience or deeper developer control.
Choose one application task, define what a useful result would look like, and test it with representative data. A focused test gives customers and developers a clear basis for deciding what to build next.
Start with one real task and a small evaluation, then expand only after the result is useful.
FAQ
Can I still use Azure Data Studio?
No, Azure Data Studio retired on February 28, 2026, and no longer receives updates, security fixes, or maintenance. Use Visual Studio Code with the MSSQL extension as the supported replacement. Existing queries, scripts, and SQL database projects do not need conversion. SQL Server Management Studio remains supported for SQL Server Agent and general SQL Server administration.
What is AI Studio and what are its features?
Microsoft Foundry is the current platform name for Microsoft’s AI development experience in Azure. It includes model access and evaluation through Foundry Models, agent building through Foundry Agent Service, and tools for adding AI functions. Teams can also use governance and monitoring features. Azure AI Studio is a former name, so older instructions may use a different interface or product wording.
When to use AI Foundry vs Copilot Studio?
Use Copilot Studio for managed low-code work with Microsoft 365 integration. Use Microsoft Foundry when developers need more control over models, orchestration, deployment, or the AI stack. The products can also work together: Copilot Studio can provide the user-facing experience, while Foundry supports advanced AI features behind it. The best choice depends on who builds and maintains each part.
Is Azure or AWS better for AI?
Neither Azure nor AWS is best for every AI workload. Compare your existing cloud commitments, the models and tools your application needs, and the development environment your team already uses. AWS offers Amazon Bedrock for managed access to multiple foundation-model providers, while Microsoft Foundry is part of Azure. A cloud’s market share does not tell you how well it fits your specific project.
Is Azure going to overtake AWS?
The Q1 2026 market snapshot does not establish whether Azure will overtake AWS. AWS held 28% of the global cloud infrastructure market, while Azure held 21% in that quarter. Those shares describe one period, not a future outcome. Cloud market positions can change, and the figures alone do not show how individual customers will choose services for later projects.
Why are people moving away from AWS?
There is no basis to assume that organizations are moving away from AWS as a general trend. A company may choose another cloud when its services, existing systems, team skills, or workload needs make that a better fit. Some organizations use more than one provider for different applications. A cloud decision is usually tied to a specific workload and its surrounding tools.
Does the US government use AWS or Azure?
Both AWS and Microsoft Azure offer public-sector cloud options. A government team must check whether a specific service meets the eligibility and requirements for its workload. Rules can differ by agency, data type, and service, so a provider’s general public-sector offering does not mean every service fits every government use. The choice depends on the stated needs of that particular workload.
