Developers currently pay for AI usage and then recover costs through subscriptions, rate limits, or pricing tiers. This creates a few problems:

1.Apps have to track per-user usage and build complex billing logic.

2.Users end up paying multiple times for the same underlying AI across different apps.

I’m experimenting with a different model: users fund their own AI usage directly, similar to prepaid mobile data.

In this system:

1.Users preload a balance 2.Each request is signed by the user 3.The cost is deducted from the user’s balance per request 4.Apps don’t pay for tokens or manage billing

I built a prototype called rNet. The protocol is live, with client libraries available (npm, Maven), and the docs are open: https://www.rnetai.org/

This is still early, and I’m mainly looking for feedback—where does this break (technically, economically, or from a UX perspective)?