JumpStart Labs is one person in Zürich shipping focused tools, one at a time. Each app is built for people who already know the field, not for people who need it explained. Surfaces stay narrow on purpose.
A daily briefing for the top one percent of AI users.
An Android-first reader for people who already know what an MoE is. Curated against the noise; ranked by signal, not engagement; opinionated about who's worth tracking.
The next one will appear here when it ships.
One app at a time. Cards take this shape: name, tagline, store badge, privacy + terms links. Nothing more.
An app that does three things excellently is more useful than a platform that does thirty things acceptably. Cut features until what remains is sharp.
If a user opens a thing every morning, the cost of a real native build is repaid inside a month. Web is the right answer for many things, but not the things people use as habit.
Onboarding tutorials are a tax on competence. The defaults should land where an expert would put them after three days of use, then get out of the way.
Most coordination overhead exists to manage scope no single person should have signed up for. Pick a smaller scope and the coordination disappears with it.