Hi, I'm Micheal.

I've been building software for over twenty-five years. Long enough to remember when "deployment" meant FTP, when "integration testing" was a Friday-afternoon problem, and when nobody used the word "DevOps", because it didn't exist yet. I started out as Build Master at the London Stock Exchange, back when Build Master meant spending hours going through CruiseControl.NET and NAnt tasks like a needle in a haystack. Before that I was a research assistant in Galway, working on telescope instrumentation with NASA, Rutherford Appleton, and the Russian 6m observatory at the BTA. I lectured for a couple of years in the Computer Science department at Galway University. Then I spent two decades building enterprise systems across Finance, international pharma, and now the energy sector.

But this isn't a CV. This is about something that's actually new.

A while back I started using AI seriously in our development work. Not the "ask ChatGPT for a snippet" kind. The integrated, agentic kind. Claude Code. OpenClaw (our own orchestration layer, a long story for another page). Full automation across builds, testing, observability and deployment, with data pipelines and telemetry built in from the start. We call this Data and AI Augmented Engineering: data-driven feedback loops, AI agents handling the routine work, and human engineers focused on the decisions that actually matter.

Every line of code is understood. Every change is reviewed and tested. We've just stopped wasting time on the parts a machine should be doing.

The result, frankly, surprised me.

My daughter Aisling is on a three-year apprenticeship through Belfast Metropolitan College, now in her second year and soon to graduate with a foundation degree in Software Development and Cloud Analytics. In the last twelve months, working alongside me and Claude Code, she built a complete bespoke CRM for an organisation that needed one. Razor Pages on .NET. Azure SQL Serverless, Azure Blob Storage, Entra ID auth. A full operational and reporting layer with custom JavaScript charting that I'd have struggled to write from scratch five years ago. The kind of system that, in 2015, would have taken a small team of mid-level developers the better part of a year. She built it at roughly a day a month on-site, while studying.

Which is the thing nobody is telling you about AI in business right now:

Data and AI augmentation isn't replacing senior engineers. It's making junior engineers extraordinarily capable, when there's a senior engineer guiding the architecture.

That's the whole insight, and it's why this consultancy exists.

ColhounTech is a small family business. I lead architecture and direction. Aisling and her brother Diarmuid (a year behind her in college, and in some ways always competing to be ahead) do most of the day-to-day building. The augmented approach lets us deliver substantially more for substantially less, without giving up the things that have always mattered: quality, understanding, testability, and someone who actually cares whether the system works in five years' time.

We've been quietly proving it on our own projects. Orbsen IQ: an AI-native CRM for coaches and consultants. VideoBatch: a cross-platform video pipeline tool. Disruptive Secrets: a research platform for X/Twitter intelligence. A handful more. Every one is shipped, in real use, and built with the same approach we'd bring to your work.

If you're a growing business with systems that need fixing, building, or rethinking, and you'd like to work with a small team that genuinely cares about the outcome, get in touch. I read every message that comes in personally.

Mícheál.