online /Dublin, IE /--:--:--
> building automation
I'm an engineer in Dublin who likes playing with automation. The fun part is figuring out which pieces actually help people: in small businesses, side projects and the small frictions of everyday life.
Below is what I'm building right now. I write down what I learn.
- projects
- 0
- videos
- 0
- posts
- 0
01 / build log
What I'm building
Small projects I use myself. Pick one to see what it is and how it helps me.
I use AI all day, in many different tools, and good ideas kept getting lost between them. So I built a map that reads my chats, notes and projects, groups them by subject, and shows what is alive and what has stalled.
Working, still improving
↳ Drag to spin it. Click a star to fly to it.
I speak instead of typing. A speech model on my own machine turns my voice into text in Portuguese or English and nothing is sent anywhere. It also lowers whatever is playing while I talk, so a video never leaks into the recording.
Working, I use it every day
↳ Press and hold the button, then let go.
I record my face and my screen as two separate files and a script assembles the cut with captions and sound. The goal: publish a video in an afternoon instead of a week.
In progress
↳ Scrub the final cut with your mouse.
A Home Assistant dashboard for what I used to check in five different apps: energy, my channels, my mail. Next to it, a 3D printer that makes small parts for the house.
In progress
↳ Hover the bars. Click the printer to restart it.
I test what is actually useful without paying a subscription for every experiment. It is where I try ideas before I write about them, like the post on running local models on a normal laptop.
Ongoing
↳ Pick a prompt and watch it answer.
02 / transmissions
Watch
More on YouTube →latest upload
03 / logs
Recent posts
-
Running local LLMs on a normal laptop in 2026: what's actually usable
An honest look at running large language models locally in 2026 — which models genuinely fit on 8, 16, and 32GB of RAM, what quantization really does, and where local still loses to the cloud.
-
n8n vs Make vs Zapier in 2026: which one I actually reach for
A practical comparison of n8n, Make, and Zapier from someone who's built real workflows in all three — how their pricing models actually differ, where self-hosting matters, and which I default to in 2026.
-
Setting up a Gmail MCP server for Claude: an honest walkthrough
What it actually takes to connect Gmail to Claude through an MCP server — the OAuth scopes that matter, the config file, the security trap most tutorials get wrong, and the state of the available servers in 2026.
-
What is the Model Context Protocol (MCP), and why I started using it
A plain-English explainer of the Model Context Protocol — what MCP actually is, how the host/client/server pieces fit together, who adopted it, and why I stopped writing one-off integrations for every tool.
-
Using Claude to draft DORA Article 30 contract annexes: what worked and where it broke
An honest look at using Claude to draft DORA Article 30 contract clauses for ICT vendors selling to EU banks. What an LLM can do well in compliance drafting, what it gets dangerously wrong, and the workflow that actually held up.