SensAI
About this project
Sep 2026 – Sep 2026 · Sole author · solo
A small chat program for a local AI model that talks to it with plain HTTP, and can force its answers into a JSON shape you define.
SensAI is a chat in the terminal with an AI model that runs on your own computer, built in the first days of an AI module. The constraint was the point: no AI library, just plain web requests.
How it works
A model has no memory, so every request carries the whole conversation. Lines starting with / are commands, and the one that matters is /format: it gives the model a template, and from then on every answer has to fit it.
The example shows why that is useful. Given a messy customer email (“my API calls have returned 401 since this morning…”), the model fills in a clean record: who wrote, which company, what kind of problem, how urgent, and a one-line summary. Software can sort and route that; it cannot do much with prose.
It is small and two days old: it does not stream answers yet, and takes one template at a time.
What I took from it
- Asking a model for a shape instead of a sentence is what makes it usable inside software.
- Starting from plain web requests shows how little a chat with a model really is.