Lab
Ideas I needed to see working.
I design the systems, controls, specifications and acceptance criteria. I use AI to implement the prototypes and verify the resulting behaviour.
Sometimes an idea stays in my head until I can see it working. I write down what the system should do and where it should stop, then work through what comes back.
These three made it far enough to keep.
Archived
Sabit
Sabit started as a month-end close tool. The AI could read the documents and suggest what might be wrong. It could not enter a number.
I stopped the product after Anthropic released finance templates that overlapped it. The Truth Gate is the part I kept.
Working demo
Tanaka
Tanaka is a small console for an idea from the essays: an AI agent should order from a menu instead of carrying the keys.
There is a working demo. One order is clean. The other has a second instruction hidden inside it.
In progress
intelligent-db
intelligent-db is an attempt to give AI agents a memory that refuses unknowns.
The fact store works. The more brain-like components did not hold up under testing. The page shows exactly which parts work, which remain partial and which are not yet built.