tinkering

A few of the things I've built. Some solve real problems, most exist purely because I got curious. Some of them work.

each one rated on the Project Questionability Meter (PQM)

wake up & breathepilates platform

My wife and a friend ran an online pilates program with 400+ classes trapped on an expensive instructor platform. I built them their own: bulk-downloaded and transcribed the everything, used local AI to generate titles and categories, tagged instructors with local facial recognition, and stood up a full site with subscriptions, streaming, and an admin panel. Took the running cost from about £100/month to about £7. Was I asked to do this? No. I was asked “please help me download my videos” - I got excited.

read the ramble → building wake up & breathe

Screenshot of the Wake Up & Breathe landing page: a practice you can return to on your own time

fridayagentic build framework

A structured, multi-agent workflow for building software with AI that produces something maintainable rather than disposable. A team of agents (planner, architect, developer, reviewer, tester) with me as the PM, plus guardrails - hooks and scripts that enforce the framework as the agents run, so the audit trail comes out of the process itself. Built, rebuilt, and still evolving. Single handedly the most valuable thing I’ve made - I used it to make this very page you’re reading! And basically everything else.

read the rambles → how the friday workflow evolved · keeping agents accountable · from driving the loop to being in the loop

local-ai home assistantwork in progress

Home Assistant runs the flat - lights, heating, blinds, media. On top of that I built a custom integration that adds a personality and persistent memory using a local model via Ollama, with a frontier model for the harder reasoning. The goal was a home that pays attention, not a speaker you shout commands at. Incredibly challenging, incredibly rewarding, still a work in progress. Sometimes a lot of fun.

read the ramble → building jarvis (minus the weaponry)

ai doorbell intercomquestionable, fun

I missed a few deliveries, so in an incredible overreaction, and possible waste of time, I tried to give my doorbell intelligence. Could it recognize a person, greet them, hear them, generate a response, speak back, all with local infrastructure? Getting audio out of the doorbell meant reverse-engineering its proprietary protocol (with a lot of help from my homie Claude). It works. Is it safe? No. Has my wife approved it for use? Also no. Was it fun? Absolutely.

read the ramble → teaching my doorbell to talk

follow-up agentit's helping

Building management companies… Am I right? Constantly emailing them to follow up on something they said they’d follow up with, only to find they’ve forgotten about it. So I created this little agent to track some threads in my gmail inbox, when the other side goes quiet, it drafts a chase using my predefined voice/tone and drops it in my drafts for review. Runs entirely on local hardware - no email content leaves my network. Built to annoy my building management company into fixing a leak, but really can annoy anybody I want (which is a lot of people). Silly? Yes. Practical? Also, yes.

read the ramble → i automated the nagging

share:share on linkedin