RAG
What's The Best Chunking Strategy For RAG Retrieval in 2026?
Start RAG retrieval with recursive or sentence-based splitting, then use evaluation results to decide whether semantic chunking is worth the cost.
6 min read
Read articleInsights
Ideas for faster, more reliable finance operations.
Articles on reducing turnaround times, preventing errors, increasing capacity, and choosing software, automation, or AI based on the problem.
Featured insight
Latest insights
voice AI
Options for Moving STT and TTS Pipelines to Australia in April 2026
Compare managed, self-hosted, colocated, and air-gapped options for moving speech-to-text and text-to-speech workloads to Australia.
6 min read
AI engineering
Should You Use LangChain When Building a New AI Chatbot or Just Write From Scratch?
Compare LangChain with provider SDKs to decide when framework abstractions help an AI chatbot and when direct code is easier to maintain.
4 min read
All insights
27 articlesRAG
What's The Best Chunking Strategy For RAG Retrieval in 2026?
Start RAG retrieval with recursive or sentence-based splitting, then use evaluation results to decide whether semantic chunking is worth the cost.
6 min read
voice AI
Options for Moving STT and TTS Pipelines to Australia in April 2026
Compare managed, self-hosted, colocated, and air-gapped options for moving speech-to-text and text-to-speech workloads to Australia.
6 min read
AI engineering
Should You Use LangChain When Building a New AI Chatbot or Just Write From Scratch?
Compare LangChain with provider SDKs to decide when framework abstractions help an AI chatbot and when direct code is easier to maintain.
4 min read
LLM evaluation
Your LLM-As-A-Judge Might Be Lying to You
Detect positional bias and ranking cycles in pairwise LLM judges, then use balanced comparisons to improve evaluator reliability.
4 min read
voice AI
What Is Actually Blocking Voice From Becoming the Default Interface?
Voice interfaces face structural limits involving memory, precision, privacy, and error recovery even as speech models improve.
7 min read
voice AI
Inference Options for Hosting a Sovereign Voice AI Stack in Australia
Compare models, GPUs, providers, latency, and costs for hosting a production sovereign voice AI inference stack in Australia.
7 min read
LLM evaluation
Stop Scoring. Start Ranking.
Learn why pairwise ranking gives more useful LLM evaluation results than numeric scoring, and how to infer relative quality.
6 min read
AI agents
CLI vs MCP: Which one should you use in 2026?
Compare CLI tools and MCP servers for AI agents across token use, context cost, reliability, security, and developer experience.
3 min read
artificial intelligence
How do you turn time with your Subject Matter Expert into a long term asset?
Turn subject matter expert feedback into reusable evaluation criteria, durable knowledge, and faster AI system improvement loops.
4 min read
AI agents
Why your agent can confidently do the wrong thing
Use deterministic routing, explicit agent skills, and evals to stop AI agents from confidently selecting the wrong workflow.
5 min read
artificial intelligence
How I stopped fighting prompts and started building feedback loops
How a generate, evaluate, and revise loop improves LLM output by turning clear acceptance criteria into useful feedback.
4 min read
artificial intelligence
Reducing AI-Slop with Logit Bias
Learn how logit bias can reduce repetitive AI writing patterns, where it fails, and how to test it without causing token loops.
8 min read
artificial intelligence
RED framework for binary AI evals. Stop debugging in the dark
Use the RED framework to make binary LLM evaluations explain their reasoning, cite evidence, and return a clear decision.
4 min read
Case Study - Hexagonal Grids for Terracotta RPG
game development
Case Study - Hexagonal Grids for Terracotta RPG
How a standalone hexagonal grid system creates procedural faction territories for a role-playing game through deterministic algorithmic expansion.
13 min read
Building Your First Evaluation System for LLM Applications
ai
Building Your First Evaluation System for LLM Applications
A practical guide to building a basic evaluation system for non-deterministic AI applications using LLMs and prompt engineering.
8 min read
Rethinking PERT, Introducing Wagn Project Tracker
estimation
Rethinking PERT, Introducing Wagn Project Tracker
Discussing the limitations of traditional PERT and introducing my personal project estimation tool, Wagn Project Tracker, which uses a certainty-based approach.
5 min read
Introducing @wagn/voice-ui: A Component Library for Voice Interfaces in React
react
Introducing @wagn/voice-ui: A Component Library for Voice Interfaces in React
Lessons from building a React voice interface library, including audio recording, streaming, and WebSocket transport for real-time conversations.
3 min read
Building Real-Time AI Chatbots - Hard Lessons Learned
chatbot
Building Real-Time AI Chatbots - Hard Lessons Learned
Lessons from building real-time AI chatbots with streaming responses, covering latency, state, error handling, and user experience.
6 min read
The Death of Frameworks
framework
The Death of Frameworks
How mature web frameworks converge on the same capabilities, and why AI may shift software value from framework expertise to problem solving.
3 min read
Samification
samification
Samification
Samification is the process where competing technologies, frameworks, or ideas gradually converge toward a single, widely accepted standard.
1 min read
Canonical Naming
canonical naming
Canonical Naming
Learn how canonical naming defines entity boundaries in Retrieval-Augmented Generation and knowledge graphs, improving retrieval and reasoning.
1 min read
Embeddings
embedding
Embeddings
Learn how vector embeddings represent meaning, support semantic search, and retrieve related content for RAG and AI applications.
2 min read
Model Context Protocol (MCP)
mcp
Model Context Protocol (MCP)
Learn how Model Context Protocol connects AI agents to external tools and data, what MCP standardises, and where it fits compared with a CLI.
2 min read
Retrieval-Augmented Generation (RAG)
retrieval-augmented generation
Retrieval-Augmented Generation (RAG)
Retrieval-Augmented Generation (RAG) is a technique that enhances AI-generated responses by first retrieving relevant information before generating an answer.
2 min read
The Value of Static Sites and My Blog Vision
static site
The Value of Static Sites and My Blog Vision
Why static sites suit durable knowledge bases: fast publishing, low operational complexity, and clear internal linking for people and search engines.
1 min read
Vibe Coding
vibe coding
Vibe Coding
Vibe coding uses natural-language instructions and LLMs to build software, changing how people prototype, debug, and learn application architecture.
1 min read
Securing Phone Calls in the Age of AI Voice Cloning
security
Securing Phone Calls in the Age of AI Voice Cloning
How AI voice cloning changes phone security, and the practical caller-verification controls that reduce impersonation and fraud risk.
3 min read