Event Recap: Building Smarter Apps with Microsoft Copilot Studio and .NET at Birmingham .NET & Xamarin User Group
I had the pleasure of speaking at the Birmingham .NET & Xamarin User Group on November 12, 2025. The talk, titled “Building Smarter Apps with Microsoft Copilot Studio and .NET,” explored the evolving landscape of AI-driven software development.
The session centered on a crucial question for developers today: “Are you steering your software development, or is AI taking the controls?” We delved into Microsoft’s AI-First Vision, the new capabilities in .NET 9, and how tools like Semantic Kernel and Copilot Studio are empowering developers to build more intelligent applications.
This is a slide-by-slide recap of the presentation titled “Building Smarter Apps with Microsoft Copilot Studio and .NET”.
Title and Introduction
Title: Building Smarter Apps with Microsoft Copilot Studio and .NET.
Central Question: The presentation revolves around the question: “Are you steering your software development, or is AI taking the controls?”.
Background: My journey included a pre-career phase (2012–2016) with UWP, reaching > 300k Users, and a career phase (2016–date) focusing on Desktop / Mobile Backend Platform.
The AI-First Vision: The core concept is Microsoft Copilot Everywhere, aiming to extend AI’s reach and embed reasoning and contextual intelligence.
Agenda and The AI-First Vision: Copilot Everywhere
Agenda (The Developer’s New Blueprint): The presentation covers six main points: The AI-First Vision, The .NET 9 AI Foundation, Semantic Kernel, Short Demo, Key SK Capabilities, and Practical Takeaways & Q&A.
The AI-First Vision: The Copilot ecosystem includes GitHub Copilot, M365 Copilot, and Copilot Studio.
M365 Copilot: Assists with daily work by offering capabilities such as recapping a meeting, drafting an email using notes and action items from a meeting, and summarizing an email thread.
M365 Copilot Capabilities and Copilot Studio
M365 Copilot (acting as a Researcher Analyst) can summarize a document (e.g., provide a bulleted list of key points from a file).
It can help tell about a topic/project, organizing the information by emails, chats, and files.
It helps with writing by generating three ways to say something or giving ideas, such as suggesting 10 compelling taglines based on a file.
It can also retrieve specifics on what a person said about a topic.
Copilot Studio is used to create or configure agents.
Copilot Control System and The .NET 9 AI Foundation
Copilots (People 1:1): Focused on Human augmentation, they function as a private, personal assistant within the Copilot Control System.
Agents (People 1:N): Are designed to automate tasks and processes, operating autonomously or connected to Copilot.
The .NET 9 AI Foundation: This layer aims for Standardization and Democratizing Access for Enterprise AI.
Democratizing Access is achieved through foundational libraries like Microsoft.Extensions.AI, Microsoft.Extensions.VectorData, and System.Numerics.Tensors.
Standardization ensures .NET features are comparable to those in Python, Java, and Go, including SDK, Sample, Documentation, and REST API.
Microsoft.Extensions.AI: The Plumbing Layer
Microsoft.Extensions.AI is described as The Plumbing Layer in .NET 9.
Functionality: It provides a Unified Interface via a new set of NuGet packages (Microsoft.Extensions.AI.*) to standardize how developers connect to and interact with various AI models.
Benefits: It promotes Simplicity by abstracting the complexity of specific model SDKs (like OpenAI or Azure AI SDKs). It also ensures Integration, plugging seamlessly into existing .NET concepts like Configuration, Dependency Injection, and the Options Pattern.
It standardizes Large Language Model (LMM) access.
Standardized LMM Connectivity
The presentation highlights several connector packages, including those for Semantic Kernel (e.g., Microsoft.SemanticKernel.Connectors.OpenAI, .MistralAI, .Google) and extensions for AI (e.g., Microsoft.Extensions.AI.Ollama).
A code example demonstrates how a developer can use a common interface (IChatClient) to get a response from an LLM, abstracting the underlying package specifics.
What is Semantic Kernel (SK)?
Semantic Kernel is defined as The AI Orchestration Layer.
Definition: It is an open-source SDK from Microsoft that allows for the seamless integration of Large Language Models (LLMs), conventional code, and memory.
Purpose: It addresses the problem that LLMs often lack the state and context of the application by bridging this gap, allowing the LLM to access and execute C# code.
Mechanism (The Bridge): SK acts as a core that runs AI prompts (Semantic Functions) and calls out to C# methods (Native Functions/Plugins).
Integration: Semantic Kernel consumes the Microsoft.Extensions.AI services for model communication, freeing SK to focus purely on orchestration logic.
SK Capability: Plugins (Native Functions)
Plugins (Native Functions) are critical capabilities that give the LLM “Superpowers”.
Concept: They are standard C# methods that the LLM treats as tools to perform real-world actions or access live data.
Mechanism: SK utilizes the LLM’s reasoning capability to determine when and how to call a specific C# function.
Example Scenarios:
Data Access: Calling a C# method like StockService.GetPrice() when prompted to “Get the latest stock price for MSFT”.
Execution: Calling an EmailService.Send() method when prompted to “Send an email to Tunde about the new project plan”.
Tool Manifest: SK automatically generates the necessary prompt instructions for the LLM based on the C# method signatures and decorators.
Integration: .NET 9 AI Stack
This section details how the various components fit together in the .NET 9 AI Stack:
LLM Core: Handles Intelligence and Text Generation (e.g., OpenAI, Gemini).
Semantic Kernel (SK): Provides Orchestration, Reasoning, and State (using Planner, Plugins, Memory).
Connector Libraries: Handle LMM Specific Interfacing.
C# Business Logic: Ensures Reliable Execution and Data Access (Native .NET Code).
It introduces the Code Example Simple AI Chat App.
Simple AI Chat App and Future Vision
The code example highlights the ease of adding multiple chat completion services using extension methods, such as AddGoogleAIGeminiChatCompletion, AddOpenAIChatCompletion, AddOllamaChatCompletion, and AddAzureOpenAIChatCompletion.
This is followed by a Chat Demo.
The Future Vision for .NET Developers: This vision includes a Shift Left on AI, meaning AI is now integrated directly into the language and framework standard, moving away from being a separate microservice.
Future Vision (Continued) and Summary
Copilots are Code: The presentation predicts that building domain-specific Copilots (Agents) will evolve into a standard architectural pattern, rather than a bespoke project.
Increased Productivity: Developers can improve productivity by focusing on defining business logic (Plugins) and data context (Memory), thereby avoiding boilerplate API calls.
The Next Generation: .NET 9 is positioned to ensure the .NET platform remains a cutting-edge choice for developing intelligent, cloud-native applications.
Summary: Key concepts reinforced are Copilot, .NET 9 & AI-First, and Semantic Kernel.
It was a fantastic evening with great questions and engagement from the community. Thanks to everyone who attended!
