Writing
What I've learned building and shipping products — for the founders and CTOs who have to make the call.
Our own neighbourhood's festival, a District programme trapped in a PDF, and a hard deadline that doesn't move. A short story about shipping something small, useful and on time.
Client work with L+R for the Global Poverty Project. A movement app where the traffic pattern is the hard part: quiet for weeks, then a stadium's worth of people at once.
Client work with L+R. An app where the output device is a skyscraper — which makes latency, turn-taking and scheduling the entire product.
Client work for Noctrix Health. A Medical-category app where the user is a trained specialist, the output is a therapy device, and "probably fine" is not a state you are allowed to ship.
Client work with L+R for Deuxmoi Media. The interesting problem is not the celebrity gossip — it is keeping an editorial voice intact once you put a filter UI in front of it.
Client work with L+R. Original audio from award-winning writers and performers — which puts the entire engineering burden on playback that never loses your place.
Client work with LabHouse. Similarity scoring across tens of thousands of photos, on the phone, fast enough that nobody puts the app down while it thinks.
Most MVP quotes are answering a different question than the one you asked. Here's what the number is actually made of.
The question isn't which is better. It's which risk you can least afford to carry right now.
The roadmap isn't slipping because engineering is slow. It's slipping because two people are using the same words to mean different things.
One of these transfers risk to your supplier. The other transfers it to you. Neither is generous.
Nobody is being lazy. The six months is designed in, and mostly by accident.
Twelve years of shipping all three. The honest version: the framework matters less than the team, right up until the moment it matters enormously.
You have two days and no context. Here's the order I look at things, and which red flags are real.
The speedup is real and it is not evenly distributed. Here's the part nobody puts in the pitch deck.
I built the product beautifully and ran out of the thing that actually mattered. Twice.
Two programmes, one year, one company that didn't make it. Four lessons that outlived all of it.