How Microsoft is Transforming Software Development with AI—One Commit at a Time

Microsoft reveals how the .NET MAUI team is leveraging GitHub Copilot as a structured collaborator, contributing real code and tests to the repo. The move signals a shift toward AI-assisted DevOps in large-scale engineering teams.

Here's a step-by-step look at how they set it up:

• Step 1: Add Copilot Instructions

  Start by including a copilot-instructions.md file to give GitHub Copilot essential context and guidance. Place it in the .github folder at the root of your repository to ensure Copilot understands your project’s structure and goals.

• Step 2: Understand Firewall Warnings

  After Copilot completes an issue, you might see a firewall warning on the pull request. This is a key security feature that helps control outbound access from the Copilot Agent.

• Step 3: Allow Trusted Domains

  To let Copilot make necessary web requests, go to Settings > Environment > Copilot in your GitHub repository. Scroll to the bottom to the Environment Variables section, and add a variable named copilot_agent_firewall_allow_list with a comma-separated list of trusted domains.

• Step 4: Run Copilot in GitHub Actions

  Copilot runs in GitHub Actions, giving you full control over how and where it operates. You can also let Copilot auto-generate a copilot-setup.yml by filing an issue.

• Step 5: (Optional) Use MCP Servers

  Set up Model Context Protocol (MCP) servers to enhance Copilot’s capabilities with specialized tools and domain context. Configure this under Settings > Copilot > Coding Agent > MCP Configuration.

Power Apps Gets a Boost with Vibe Coding and Enterprise-Grade Trust

Microsoft is ushering in a new era of app development with the introduction of Vibe Coding in Power Apps. This innovative approach blends generative AI with structured collaboration, empowering makers to build with more creativity and confidence. Backed by enterprise-grade trust, the update reinforces Microsoft’s vision of secure, AI-powered low-code development.

Here’s what you need to know about vibe coding:

• Start with a vision

  Describe the app you want using a simple prompt. Agents help you shape user stories, requirements, processes, and architecture in one seamless flow.

• Bring in your data

  Upload sketches, images, or select your Dataverse tables — and let your data guide the build.

• Generate and refine code instantly

  Watch your app come to life instantly. No code required — but full customization and control remain in your hands.

• Enjoy the best of both worlds

  Vibe coding bridges no-code and pro-code, giving you transparency, flexibility, and an iterative design experience.

• Expanded availability

  Preview access is now available across North America — with global rollout coming soon, so more makers can start building with vibe coding.

Microsoft to Retire Cognitive Services & Azure ML for Power BI Dataflows by Sept 15, 2025

Microsoft has announced that its Cognitive Services and Azure Machine Learning integration within Power BI dataflows will be officially retired effective September 15, 2025. Starting August 11, 2025, users will no longer be able to create new AI models via Cognitive Services, and the current user interface entry point will be removed. After the retirement date, existing AI‑powered dataflows will cease refreshing and will no longer be supported, urging customers to transition to alternative solutions before that deadline.

Here’s what you need to know:

• What’s changing?

  Starting August 11, 2025, users can no longer create new models using Cognitive Services in Power BI. Existing models will still refresh until September 15, after which support ends entirely.

• Why the change?

  While Cognitive Services and Azure ML made it easy to build predictive models in Power BI, users asked for more flexibility and control over their ML workflows.

• What’s replacing it?

  Enter AutoML in Microsoft Fabric’s Synapse Data Science — a modern, powerful, and highly customizable solution for automating machine learning.

• What can it do?

  It enables you to automate ML workflows, visualize outcomes, and train models using scalable tools — all within one unified experience.

• Use cases include:

  Customer segmentation, churn prediction, demand forecasting, and embedding predictions directly into your Power BI dashboards — all from a unified UI.

Power Apps Enhances Visual Collaboration Experience

Power Apps is taking another step forward in helping makers turn data into action with greater efficiency and clarity. A recent update introduces improvements designed to enhance how users collaborate around data-driven insights. These enhancements aim to make analytical exploration more seamless and reusable within app experiences.

Here’s what’s new:

• Save and reuse visualizations

  Your insights now stay with you. Save charts to revisit, apply them across records, and align your team around the same story—no more losing work when you move on.

• A refreshed look for all charts

  Legacy or Copilot-generated—your charts now share a consistent, modern design. Enjoy improved readability and styling, whether built manually or with AI.

Ask Mode vs. Agent Mode: What’s Best for Your Workflow?

With GitHub Copilot becoming an essential part of modern development, .NET developers now have more control over how it supports their workflow. A new guide from Microsoft breaks down the difference between Ask Mode and Agent Mode, helping you choose the right Copilot experience based on your development needs.

Here’s a quick guide to help you decide between Ask Mode and Agent Mode—so you can boost productivity and solve problems faster as a .NET developer:

• Understanding Ask Mode

  Ask Mode is ideal for quick, focused support—Copilot responds based on the context you provide, without accessing your project files.

• Ask Mode is great for:

  Explaining C#/.NET concepts, providing code snippets, summarizing documentation, and sharing best practices or design patterns.

• Understanding Agent Mode

  Agent Mode works like an intelligent assistant embedded in your solution—it can read and reason about your actual codebase, perform edits, and automate tasks.

• Agent Mode is perfect for:

  Refactoring code, generating tests, fixing bugs, performing code analysis, or automating repetitive tasks using your real project context.

Bring Voice Intelligence to Your Apps with .NET MAUI

The latest innovation in .NET MAUI brings multimodal voice intelligence to your fingertips, enabling developers to create richer, more interactive app experiences. With native voice input and output capabilities, your apps can now engage users in entirely new ways—combining audio, UI, and AI.

Here’s how to do it:

Step 1: Add the Plugin.Maui.Audio NuGet package to request microphone permissions, capture audio streams, and enable playback functionality.

Step 2: In MauiProgram.cs, configure recording settings and register IAudioService in the service container. Don’t forget to check the plugin’s documentation for any additional configuration steps.

Step 3: In your VoicePage, allow the user to tap a microphone button to start and stop recording. Implement this interaction using a simple if-else logic.

Step 4: With the audio stream captured, begin transcribing and processing the input.

Step 5: Use Microsoft.Extensions.AI with OpenAI (or your preferred service) to transcribe the audio.

Step 6: Once transcribed, analyze the text and return relevant insights—like projects or tasks—using your AI client, handled in the ExtractTaskAsync method.

Developers can now tailor GitHub Copilot AI chat responses with custom instructions

Microsoft’s .NET team has introduced a way for developers to embed “custom instructions” files in their repositories, enabling AI to adapt to project-specific workflows, code conventions, and review guidelines . Released six days ago, the update lets teams centralize their preferences—such as naming patterns, error‑handling styles, and preferred frameworks—within a simple markdown file. This ensures Copilot Chat starts each interaction already primed with the right context for the project.

Here’s how to do it:

• Step 1: Create a .github folder at the root of your solution.

• Step 2: Add a copilot-instructions.md file inside that folder.

• Step 3: Write your custom instructions in Markdown (there’s a great community repo for inspiration).

• Step 4: Open GitHub Copilot in Agent mode, load your project, and ask away!

Once Agent mode runs, Copilot uses your instructions to tailor its responses. Give it a try and take control of your AI assistant.

Power BI Evolves Again in July 2025 Release

Power BI continues its steady evolution with the July 2025 update, signaling Microsoft’s commitment to delivering a more intelligent, intuitive, and deeply integrated analytics experience. This month’s release reinforces Power BI’s role not just as a reporting tool, but as a collaborative, AI-driven platform for decision-making across organizations. With a focus on user experience, governance, and smarter workflows, this update lays the groundwork for a more connected data future.

Here are some of its latest updates:

• Copilot and AI
  Tenant admins can now control Copilot’s default search behavior—limiting it to content explicitly marked Prepped for use with AI. Smarter access, safer insights.

• Reporting
  A new left-hand navigation menu in the Power BI app for Teams makes it easier to discover and access your reports without losing momentum.

• Modeling
  You can now build Direct Lake semantic models in Power BI Desktop directly from SQL and Mirrored databases in Microsoft Fabric—unlocking real-time insights at scale.

• Data Connectivity
  The Snowflake connector 2.0 is now generally available, enabling faster retrieval of large datasets without unnecessary overhead.

• Mobile
  Power BI Mobile’s Home screen is now reorganized to prioritize your most-used content—bringing insights to your fingertips even faster.

View the full list here.

Microsoft Extends “Least Active Routing” to Messaging Channels in Dynamics 365 Contact Center

Microsoft has rolled out a key enhancement to Dynamics 365 Customer Service, expanding its intelligent routing capabilities. The update brings a more refined approach to handling messaging channels, building on existing innovations in service distribution. This marks another step in Microsoft’s ongoing commitment to improving agent efficiency and customer experience.

• What is Least Active Routing?

  It’s a smart way to fairly distribute tasks by assigning conversations based on the last capacity release time of service reps — ensuring balanced workloads and better service delivery.

• Why use least active routing?

  It ensures adequate rest between tasks, better utilization, improved customer satisfaction, as well as better customer service representation.

• What’s new?

  It now factors in concurrent workloads and considers voice and messaging activity when assigning new tasks.

• Use Case 1:

  An agent with fewer ongoing chats will be assigned the next one to balance workloads.

• Use Case 2:

  If agents have equal workloads, the one who’s been idle longest and has the least capacity will receive the next chat or call.