Why Does AI Need So Many Data Centres?
One agent task can mean 20 model calls. Multiply that by everyone.
Read on Substack →Learn
Practical writing for .NET and Azure teams working out where AI fits, from Learn to Build AI.
To learn to build AI as a .NET developer, build one small agent in C# that calls one tool, then add MCP for tools and A2A for agent-to-agent calls. These articles are grouped by theme, each with working code.
Last updated
Latest
Building AI with .NET and Azure
Agents, MCP, A2A and Foundry in C#, with working code, and how to add AI to the systems you already have.
Your Azure Functions Are Already MCP Tools - Here’s the Last Step
One NuGet package and a few attributes turn your existing Azure Functions into tools any MCP client can call.
Read on Substack →Build Your First Agent Workflow in Microsoft Foundry - with a C# Client
Foundry's workflow builder chains agents without orchestration code, and the docs only show Python, so here is the C# call.
Read on Substack →Agent Skills in .NET and the Microsoft Agent Framework
One agent with its capabilities in composable, versioned markdown files instead of one giant prompt or five agents.
Read on Substack →Building a 2-Agent A2A System from Scratch in C#
An orchestrator and a code review specialist wired together with capability-driven discovery.
Read on Substack →Agent2Agent (A2A): The Other Protocol .NET Architects Need to Know
MCP is how an agent uses a tool, A2A is how it talks to another agent, and how to design for both.
Read on Substack →Building CodeGuard AI: A Multi-Agent Code Review System
A multi-agent C# code reviewer with RAG, MCP filesystem access and clean architecture on .NET 8.
Read on Substack →How to Add AI to Your Existing .NET Application Without a Rewrite
When the brief is "add AI by end of week", you can start small inside the app you already have.
Read on Substack →I Built an AI Homework Generator App
A small, live app built around one prompt, and what that taught me about shipping AI features.
Read on Substack →Working with AI
How to get real work done with AI tools and agents, and what is going on under the hood.
The Three Levels of Working With AI
Most people stop at a chat tab, and each of the next two levels changes what you can get done.
Read on Substack →Build Transformative AI Assistants with Claude Code Workspaces
Six months of using workspaces with MCP tools and commands as a knowledge worker's second brain.
Read on Substack →How Agent Skill Execution Actually Works
A walkthrough of the reasoning loop behind skills, so you can predict what your agent will do.
Read on Substack →Learning paths and career
What AI means for your career as a developer, and how to learn it without burning out.
AI Opened the Floodgates. Now We Need More Developers Than Ever.
AI isn't replacing developers, it's exposing how much software the world still needs and how few of us there ever were.
Read on Substack →Where Do Senior Developers Come From Now?
Seniors learned on the work AI absorbed first, so where does the next generation come from?
Read on Substack →Why I'm Diving Deep into AI Agents
The few moments in my career when I knew the job was about to change, and why this is one of them.
Read on Substack →Adopting AI in teams
Bringing AI into a team, what works, what to expect, and the mess that follows a mandate with no plan.
The AI Mandate and the Mess It’s Making
"Uses AI" is now a job-spec filter and a top-down order, and the gap between the mandate and the day-to-day is where things go wrong.
Read on Substack →Get new posts by email
Free, no spam, and you can leave whenever you like.