# How we work

How we work, in four rules. Each one is something we already do in our own products.

We do not build ad tracking or data resale into products.

We build native for Apple platforms and for the web. AI agents work alongside us, on [a platform we wrote](/work/wapps-platform/) for that.

## Write the decision down, then build it.

Every feature starts as a dated spec and a plan in the repository, and ends with a note on what was built.

## Test the rules, not only the screens.

In Emlak Radar, every rule has a test, and every database query is tested against the database's real limits.

## Collect only what the product needs.

Couch Remote collects no data. Emlak Radar turns speech into text on the phone and keeps no audio.

## Make every app work on sample data.

Our iPhone apps open on sample data without our servers, for reviewers and for screenshots.

## Brief.

You get a written brief of what we will build and what we will not.

## Plan.

You get a scoped plan with the order of work and the first release.

## Build.

You get a written update every week. It says what shipped, what is blocked and what is next.

## Release.

We ship to the store or the web, and we offer to maintain it.