Deterministic Voice Forms with Blazor and Local LLMs

A Ten Commandments–Compliant Architecture In which we build a voice-to-form system that doesn't hand the keys to the AI kingdom Here's the thing about voice interfaces: everyone wants them, nobody trusts them. And they're right not to. Most voice forms fail…

GraphRAG Part 2: Minimum Viable GraphRAG

In Part 1, we explored why GraphRAG matters. Now let's build a minimum viable GraphRAG with three extraction modes: Mode LLM Calls Best For Heuristic (default) 0 per chunk Fast indexing, structured markdown Hybrid 1 per document Balance of speed and quality…

GraphRAG: Why Vector Search Breaks Down at the Corpus Level

Your RAG system is great at "needle" questions: retrieve a few relevant chunks and synthesise an answer. It struggles with two common query types: Sensemaking: "What are the main themes across this corpus?" Connective: "How does X relate to Y across different…

Ephemeral Signals - Turning Atoms into a Sensing Network

A tiny primitive that turns concurrent work into a coordinated, adaptive system. "The Ephemeral Signals Pattern" In Part 1 we built ephemeral execution - bounded, private, self-cleaning async workflows. In Part 2 we turned that into a reusable library with…

Building a Reusable Ephemeral Execution Library

In Part 1: Fire and Don't Quite Forget, we explored the theory behind ephemeral execution - bounded, private, debuggable async workflows that remember just enough to be useful and then evaporate. This article turns that pattern into a reusable library you can…

Fire and Don't Quite Forget - Ephemeral Execution Without State

Most async systems either remember too much (logs, queues, persisted junk you never wanted)… or they remember nothing at all (fire-and-forget black holes that vanish the moment something goes wrong). This article introduces something different: A pattern where…

Learning LRUs - When Exceeding Capacity Makes Your System Better

Most systems degrade when overloaded. Memory fills up, queries slow down, users complain, servers crash. A few unusual ones get better. This article shows how an LRU-based behavioural memory becomes self-optimising when it hits capacity - and how this pattern…

StyloBot: Fighting Back Against Scrapers (Part 1)

Scrapers are about to start using AI to mimic real users - so I built a bot detector that learns, adapts, and fights back. Read Part 2: How Bots Got Smarter - The New Frontier in Bot Detection 👉 See It Live: StyloBot.net - This is the real production system…

Load Testing ASP.NET Core Applications with k6: Practical Implementation

Now that you understand the fundamentals of k6 and performance testing, it's time to put that knowledge into practice. This article walks you through writing real tests, integrating them into CI/CD pipelines, and using profiling tools to identify and fix…

Load Testing ASP.NET Core Applications with k6: Introduction

Your application works perfectly on your laptop. Unit tests pass. Integration tests pass. You deploy to production, and suddenly everything grinds to a halt. Five hundred real users hit your homepage simultaneously, and your server starts returning 503 errors.…