Today, we’re excited to share our public roadmap, which outlines the next steps in evolving Visual Studio with AI-powered agentic experiences. With every month, we aim to deliver smarter, faster, and more intuitive tools that enhance your coding experience.
Disclaimer: The items outlined here represent ongoing work for the month. They are not commitments or guarantees for delivery within the current month.
With that said, here is what we are working on!
New Modes for Debugging and Profiling:
We’re streamlining how you find and switch between modes and making sure both built-in and extension-provided modes can handle more complex workflows.
Two new agent-powered tools are in progress:
Supporting Customization via Modes
We are making it seamless to integrate your workflows across your development stack into Visual Studio, tailoring them to you and your organization.
Agent Mode/Chat:
We’ve been listening to your feedback on Agent Mode and Chat, and we’re making some big improvements.
Faster performance
Better visibility
Expanding Agent Mode functionality
Smarter context retention
More control
Model Context Proctol (MCP)
We want you to bring your entire development stack into Visual Studio, supported by the same security, governance, and trust you expect from our product. To make this possible, we’re expanding support for these capabilities and making them easier to use.
Models
We’re committed to giving you access to the latest models, and in Visual Studio we carefully evaluate them to make sure you get the best possible experience. With the recent introduction of Bring Your Own Key, your choice of models is expanding even further.
GitHub Copilot native support:
Bring Your Own Key support:
Azure Agentic DevOps
We’re bringing the power of agentic workflows and Azure-native capabilities directly into Visual Studio, so you can build, deploy, and manage your applications with the same security, governance, and trust you expect from our platform. The public preview is now available in Visual Studio 2022, featuring an automatic installation of Azure MCP for a seamless start.
Here’s what’s coming next:
- 1P Azure MCP installation in Visual Studio
- Publishing and CI/CD (GitHub Actions & Azure DevOps) support to Azure App Service for .NET applications
- Custom modes tailored for Azure developers
Editor improvements
Copilot-powered code completions and suggestions in your editor are triggered effortlessly and can accelerate your daily programming. We want to deliver a smooth and intuitive experience for you when interacting with these suggestions that minimize potential conflicts.
To make Visual Studio a truly AI-integrated IDE, we want to ensure that Copilot is seamlessly available at every step of your development workflow—not just for writing code, but also for searching, fixing errors, writing unit tests, and even committing and pushing your changes.
We’re excited for you to try these new experiences soon. If you have feedback, post it in the developer community ticket linked above. For other ideas or suggestions, drop a comment below or create a new ticket—our team reviews them all.
Thanks 😊
My feedback is that my Visual Studio customers are running in air-gapped data center environments and cannot use public LLMs. I ship a tailored hardware/software solution which runs LLMs locally via Ollama. My customers would like to configure Visual Studio to work with local LLMs.
Microsoft has low latency high quality models in the regards to the phi4 series of models which seems like it would be perfect for this task.
Ugh. I keep forgetting if you have two copies of Visual Studio open and are using Agent on two codebases at the same time it basically gets confused and splices the code from each repo into the other one, essentially scrambling two repos. I’ve fallen for this twice now wasting dozens of hours pulling code back out of copilot logs😭 I’ve tried to ping you on Twitter, but I don’t think you check there.
I sincerely want to not come across as rude, but I think my adverse experience with your product might be useful to you.
It would be really helpful if you could allow Pro and Pro+ subscribers to perform additional paid premium requests once they hit the 100% limit.
I hit my limit and tried to remove the budget limit so that I could pay for additional requests. I spent hours trying, until I realised it's not implemented yet - yet its is marketed everywhere that you can "Purchase additional premium requests at $0.04/request".
Feels a little sneaky, is all. Consider not...
Thanks for the feedback! To manage your Copilot subscription, open the Copilot badge in Visual Studio. From there, you’ll see an option that redirects you to GitHub.com/Settings, where you can adjust your subscription settings (including switching to pay-as-you-go).
In the future, we’d like to bring more of these controls directly into Visual Studio, but it’s not currently on the roadmap. If you’d like to see this prioritized, please file a DevCom ticket https://developercommunity.visualstudio.com/VisualStudio.
You guys seem busy for the next few days 🙂
Looking forward to the editor enhancements. Suggestion: add the most reasonable Copilot suggestions to the Context menu in the editor, not only a “Ask Copilot”.
While at it, check why code completion and suggestions are not working while debugging and editing code.
Hey Andreas! Thanks for your feedback 🙂
– Copilot in context menu: What are some actions you’d like to see to be added to the context menu? Could you leave your ideas in this devcom ticket? https://developercommunity.visualstudio.com/t/Add-More-Copilot-Actions-in-the-Context/10951154 Would love to learn from you
– For code completions and suggestions, they are off by default in the debugging mode because previous customer feedback we got. But you can turn them on Tools > Options > IntelliCode > Advanced and turn off Disable suggestions while debugging.
Well, but personally Codex + ChatGPT works really wonderfully, even on Windows with WLS.
With External Tools:
Title: Codex
Command: C:\Program Files\WSL\wls.exe
Arguments: –distribution-id {Guid} –cd “$(SolutionDir)”
Initial directory: $(SolutionDir)