Experimenting with an ISP-style billing model for AI usage

Right now:

1. Developers have to track each user's usage, build complex pricing systems to stay profitable, and handle upfront AI token costs.

2. Users pay multiple times for the same AI across different apps.

rNet is an attempt to flip that into something closer to an ISP model, where users fund their own usage directly.

This is still an early proof of concept. The protocol is live, libraries are published on npm and Maven Central, and the docs are open.

I’m mainly interested in feedback on where this breaks — technically or conceptually. url:- https://www.rnetai.org/