Insights.
What we've learned building AI and software that people use every day: retrieval, evals, agents, agentic engineering and the platforms around them.
What is an MCP server? Uses and examples
An MCP server gives AI assistants and agents a standard way to use one of your systems. How one works, what companies use them for, how they compare with an API and what to secure before you connect one.
9 min read
All insights
What is RAG? How it works, with examples
What RAG is in plain words: retrieval-augmented generation lets a language model answer questions from your own documents and cite them. How it works, four examples, how it compares with fine-tuning and long context, and what goes wrong.
9 min read
Vibe coding vs agentic engineering explained
Vibe coding means asking an AI tool for software and accepting the code without reading it. Where that works, what breaks once real users and data arrive, how agentic engineering differs, and what to do if your business already runs on a vibe-coded app.
9 min read
AI governance: a practical framework
AI governance is the set of rules, roles and checks that decide how your organisation uses AI. This article covers what a framework includes, how to set one up, who owns what, where the EU AI Act fits and what changes once agents take actions.
9 min read
EU AI Act summary for businesses using AI
The EU AI Act sets rules by how AI is used and by your role as provider or deployer. This summary covers the risk levels, the dates as of 16 September 2026, deployer duties, UK scope and penalties.
10 min read
How to automate your business processes
How to automate business processes one at a time: pick a repetitive process, measure it, decide between rules and AI for each step, and run it alongside your team before it takes over.
10 min read
API integrations explained, with examples
An API integration lets two systems exchange data without anyone re-typing it. How one works, examples from finance and operations, the ways to get one built, what can go wrong and what drives the cost.
10 min read
What is an AI agent? Examples for businesses
An AI agent is software that carries out a task in several steps across your systems, within limits you set. What that means in practice, how it differs from a chatbot, three examples and when to build one yourself.
11 min read
How to build a RAG chatbot people can trust: a practical guide
What it takes to turn a promising RAG experiment into a chatbot people trust: ingestion, hybrid search, citations, permissions and evals.
6 min read
LLM evals: testing AI before every release
A practical approach to LLM evals: build a test set from real cases, combine code checks with model grading, and block releases that regress.
5 min read
Agentic engineering: how we ship SaaS with Claude Code and Codex
How experienced developers use coding agents to build SaaS while keeping control of architecture, quality and security.
4 min read
RAG vs fine-tuning: which one do you need?
RAG gives a model your knowledge at answer time; fine-tuning shapes its behaviour. How to choose, when to combine them and what each costs.
4 min read
Human-in-the-loop AI agents: design patterns for real operations
How we keep a human in the loop when AI agents run on live systems: approval checkpoints, confidence thresholds, tool permissions, audit trails and fallbacks.
5 min read
When custom software beats off-the-shelf
A practical way to decide between off-the-shelf software and custom development, covering total cost of ownership, integrations and risk.
5 min read
Embeddings explained: a practical guide for product teams
How embeddings power semantic search, how to choose a model and a vector store, and how to measure whether your results improved.
5 min read
PDF data extraction with LLMs: invoices and forms
How to use LLMs for PDF data extraction: turning invoices and forms into reliable structured data with schemas, validation and a review queue for the cases that need a person.
5 min read
How to evaluate RAG: retrieval metrics, faithfulness and golden sets
How to do RAG evaluation properly: separate retrieval from answers, check faithfulness claim by claim and build a golden set you can trust.
4 min read