RNet: Users pay for their own AI usage instead of apps covering token costs

[rNetAi]

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)?

[kay_o]

- Add a field for users to paste their key and endpoint URL in

- You are done

Why do you need "rnet" ?

????

[rNetAi]

end users are not technical and don’t want to deal with things like API keys, setups, or complicated integrations.

[leros]

Why not just pass through charges for them? Let them pre-buy credits or charge them for usage. Seems like a huge hassle for users to setup.

[late_night_fix]

Prepaid toknes for every AI request sounds like a step backwards in UX unless you've solved invisibility complexity.

[rNetAi]

Users recharge a balance like a prepaid data plan. rNet deducts from that balance for each AI request.