About
The lab, in longer form
WizusLabs is a small studio that makes consumer software. The staff is mostly not human. Five crews of AI agents write the code, draw the interfaces, score the feedback sounds, run the tests, and push the builds to the stores. One person — the human in the loop — decides what the crews work on, reads the diffs, and signs off on every release. If you have downloaded one of our apps, a machine wrote it and a human said yes to shipping it.
Origin
The lab began as a personal project: a single developer asking whether a coordinated team of specialist AI agents could actually ship a real, maintained, store-reviewed consumer app — not a demo, not a thread, not a screenshot. The first answer came when NeuralSpark reached a store listing. The second came when Sudoku by WizusLabs went to review. The third came when Iron Swarm reached the stores. The lab exists because the first experiment worked, and because it was more interesting to keep going than to write a blog post about it.
The model
Twenty-one specialist agents are grouped into five crews, each with a narrow mandate:
- Wizus Bridge reads the world, picks the target, and charts the route. Research, planning, architecture, and economy design live here.
- Wizus Atelier decides what the thing should feel like. Creative direction, visual design, game feel, audio, and illustration live here.
- Wizus Workshop turns specifications into software. Implementation, persistence, data modelling, and localization live here.
- Wizus Foundry handles the pipes, the languages, and the written record. Infrastructure, CI, translation, and documentation live here.
- Wizus Gate tests, signs, and ships. Quality assurance, release clearance, and the build pipeline live here. Nothing leaves without a seal from this crew.
One human sits above all of them. They set the direction, adjudicate disputes, and carry legal responsibility for what lands in a store. They answer support email at a public address. They do not have a photo on this site on purpose — the apps and the log are the face of the lab.
Why AI
The honest answer is: because it is possible now, and because the work is more interesting this way. A solo developer coordinating twenty-one agents can hold more in their head at once than a solo developer coordinating themselves. The agents do the kinds of work that reward consistency and stamina — writing tests, localizing strings, auditing for accessibility, keeping a changelog honest. The human does the kinds of work that reward judgment — picking what to build, noticing when something feels off, deciding when a feature is actually done.
We do not think this model replaces a conventional team. It is a different shape. It produces software that, so far, the stores accept and the players keep.
What we don't automate
Some things the human still does by hand. Reading every release note before it ships. Replying to real emails from real users. Deciding when a bug is a ship-blocker and when it is a known issue. Writing this page. Signing the manifesto.
There are also things the lab will not do, even though it could. We do not write fake reviews. We do not buy install traffic. We do not publish metrics we cannot source. We do not costume the agents as humans. They are agents; we name them; the page you are reading says so.
Where we're going
The near term is direct: keep Iron Swarm, NeuralSpark and Sudoku healthy, translate the backlog into more languages, improve the log so it generates itself, and keep the site plain enough that it will still work in ten years. If a fourth app makes sense, we will add it. If it does not, we will not.
If you want to follow the work, the shipping log is the honest place to look. It updates when something actually ships. That is the whole editorial policy.