Topics
Browse all 24 topics covered on the blog.
.NET 60 guides
Everything here is written for the .NET runtime rather than translated from a Python tutorial: C# examples, the real package names, and the failure modes that only appear once an agent runs inside ASP.NET Core or a container. If you are starting from nothing, Microsoft.Extensions.AI is the abstraction the rest of the stack now sits on.
AI Agents 36 guides
An agent is a loop. The model asks for a tool, your code runs it, the result goes back, and it decides again. These guides cover what that loop costs per turn, where it fails quietly, and the controls worth putting around it before it can touch anything that matters.
C# 28 guides
Articles where the C# is the point rather than an illustration — tool methods, typed responses, cancellation, disposal — including the places the SDK surface has moved recently enough that search results still show you the old shape.
Architecture 20 guides
The decisions taken once that are expensive to revisit: where conversation state lives, how many round trips a single turn really costs, which side of a process boundary a tool belongs on, and what adding a second tenant does to all of it.
Microsoft Agent Framework 14 guides
The framework that absorbed Semantic Kernel and became Microsoft's agent story, GA since April 2026. Sessions, tools, handoffs, the harness and the hosted runtime — with an honest account of what still frustrates.
Tutorial 10 guides
Start-to-finish builds you can follow with an editor open. Each one ends somewhere that actually runs, and then tells you what breaks first when real traffic arrives.
RAG 8 guides
Retrieval-augmented generation treated as an engineering problem rather than a diagram: how you split documents, what vector similarity does and does not capture, and how to tell a retrieval failure apart from a generation one before you start tuning prompts.
Security 7 guides
Prompt injection, tool authorisation, tenant isolation and redaction. The recurring theme is that a control the model can talk its way past was never a control, and that the boundary has to live somewhere the model cannot reach.
Azure 6 guides
Azure OpenAI, Container Apps, Functions and Foundry Agent Service — with the quota model, the ingress timeouts and the scaling behaviour that only become visible after the first deployment.
Embeddings 6 guides
Vectors, similarity and the stores that hold them. Including the failures that raise no exception at all: two embedding models mixed in one index, chunks too broad to match anything, and queries that were never semantic to begin with.
MCP 6 guides
The Model Context Protocol: building a server in C#, consuming somebody else's, putting OAuth in front of one, and deciding whether a protocol boundary is worth the latency it adds to every call in the loop.
Deployment 5 guides
Getting a .NET agent onto Container Apps, GKE, Azure Functions or a local Aspire stack, and the platform defaults that will cut your longest requests in half without telling you.
Production 5 guides
Cost control, rate limiting, retries, human approvals and multi-tenancy — the work that sits between a demo that impresses a room and a service you are willing to be paged for.
OpenAI 4 guides
Calling the OpenAI API from C# directly: chat completions, streaming, function calling, counting tokens before you send them, and reading the parts of the response that most tutorials never mention.
Cost 3 guides
What a completed task costs, rather than what a token costs. Caching, counting, and the input-side growth that nobody notices until it arrives on an invoice several weeks later.
Model Providers 3 guides
Choosing between Anthropic, Google and OpenAI from .NET — and what genuinely differs once a common abstraction has flattened the client code to the point where swapping one for another compiles.
Semantic Kernel 3 guides
Where Semantic Kernel stands now that the Agent Framework has absorbed it, what the two share underneath, and how to move an existing codebase across without booking a rewrite weekend.
Smaller topics