Our Stack, Audited

The annual cull

Once a year we list every tool the studio pays for, print the list, and defend each line item out loud. If nobody can remember the last time a tool earned its subscription, it goes. This year twelve survived. Four did not, including one we had been paying for out of pure nostalgia.

The survivors share a trait: they are boring in the best way. Fast to open, obvious to use, quiet about updates. The tools that got cut were the ones that kept asking for attention with onboarding tours, changelogs, and notification badges. Attention is the studio’s only real currency, and we stopped spending it on software.

The shortlist, with reasons

Design lives in two tools, on purpose: one for thinking in systems, one for making things move, and a hard rule against letting either do the other’s job. Every time we have blurred that line, the files got heavy and the decisions got slow. Tools are opinions with keyboard shortcuts; owning fewer opinions keeps the work coherent.

Handoff is where subscriptions go to die. We replaced three specialised platforms with one shared space and a naming convention, and clients stopped asking where things live. The convention did more than the software: predictable names beat clever features in every audit we have run.

Operations stayed deliberately boring. One place for words, one for money, one for time. The test we apply is cold-start speed: if a new collaborator cannot be productive in a tool within an hour, the tool is doing too much.

What almost made it

Two candidates fought hard for a slot this year. We cut an AI notetaker that summarised every call beautifully and made us stop listening. We cut a whiteboard tool with genuinely lovely physics that had quietly become a second, worse file format for decisions, and rebuilt its best boards as plain documents.

The lesson repeats every audit. Software that produces artefacts must answer a single question: where does this live in six months? If the answer is a screenshot in a chat thread, the tool is renting attention, not saving it.

Next year’s list starts from zero again. Incumbency is not a feature.

The one we cannot quit

A tiny window-management utility that nobody outside the studio has heard of. It saves maybe four seconds an hour. Over a year, that is two full working days returned to us: the cheapest hire we have ever made.