Many people want the same AI answer or tool and each pay for it alone.
Problem statement
Strangers who want the same answer pay for it separately. Sharing logins to split the cost can break provider terms and expose keys. How might we let them fund one AI run together and stay in control of it?
What sparked it
Three public reports in the sources framed the idea. They cover open-source devs fighting AI crawlers, Meta delaying new business terms after an outcry, and a Disney+ price jump.
Why this approach
A browser extension opens on any page with an open question. It turns the question into a capped, public AI run that backers steer.
Hypothesis tested
We believed that strangers will fund one capped AI run if they approve each milestone. If it holds, people will pledge small sums to a stranger's question.
How to try it
Open the Prototype tab and pick a use case card.
Click a chip on the framed page. Merge the suggestion cards and approve an objective.
Drag the pledge slider, press and hold to pledge, and simulate backers until the jar is full.
Start the run, read the log, and vote on each milestone.
What this does not do
It does not move money or call a real AI provider.
The prompt merge is scripted, and the vote counts are illustrative.
It reads no privacy policy. Each cell only says what the agent would read.
Notes
Who would pay: backers, through a 3 to 5% fee on funded runs (assumption), and institutions that want verified results.
First customer: a digital-rights group that wants a public comparison of AI privacy policies.
Closest competitor: FablePool, which pools credits behind software projects.
Why Mozilla: the browser is a user-controlled surface, so people can act together on any page.
Feedback question: would you pledge $2 to a stranger's question if you could approve each milestone?