The prompt became the unit of work. FablePool just made it the unit of price.
A site put a live price tag on raw instructions. The public ledger shows what the market thinks they're worth.
A new site hit the Hacker News front page today. FablePool. 311 points, 174 comments. The pitch is one line: pool money behind a big prompt, and an AI agent attempts the build, in public.
The mechanic is clean. You write one ambitious instruction. An AI planner reads it and sets a funding target, with every project totaling at least $100. Strangers chip in anything from $0.25. Once it funds, an agent runs the build milestone by milestone, and every credit it spends lands on a public ledger anyone can read.
That ledger is the part worth reading. Not the pitch.
What the ledger actually shows
Look at the open projects.
"Build a completely greenroom, open source AWS." $3.25 raised against a $516 target. "Tech Messiah." $4.00 of $740. "Make Fable 6." $1.00 of $205. "Port Notepad++ to MacOS." $1.00 of $365. Most of the board sits at a dollar or two against targets in the hundreds.
One project shows completed: an open-source Turbopuffer-style search database, $145 raised of a $339 estimate. One.
So the front page sells a market for funding prompts. The ledger underneath shows a market that has mostly funded nothing.
The top comment on Hacker News is "$516 for a complete AWS replacement, I'm in lol." It got upvoted because everyone read the same gap at the same time. The compute estimate is real. The deliverable is fantasy. The laugh is the distance between them.
Two prices are being confused
FablePool's planner prices one thing precisely: what it costs to run the prompt. One commenter watched the demo build come in estimated at $0.35, actually cost $0.52, charged $0.55. The estimator is close on compute. It knows what the tokens cost.
But that is not the number that decides whether a project funds.
The value of the thing built is the other price, and the planner does not set it. The crowd does. Nobody funds the AWS replacement at $516 because $516 of compute does not produce an AWS replacement. The estimate measures the build run. It says nothing about whether the build is real at the end.
People read FablePool as crowdfunding for software. That is the visible layer. Underneath, it is a public auction for prompts, and the bids are telling you what a prompt is worth on its own. About a dollar.
That is the finding. A site finally put a live price tag on raw instructions, and the market priced them at near zero.
What this means for the builder in the middle
The instinct, after a year of building with AI, is to read this as a place to get your idea funded. Write the big prompt, watch strangers back it, let the agent ship it while you sleep.
The ledger says otherwise. The prompt was never the scarce thing.
Writing "build an open-source AWS" costs nothing now. Anyone can type the most ambitious instruction on the board. That is exactly why the board is full of them and empty of money. What strangers will not pay $200 for, sight unseen, is the part that is still hard: scoping the instruction down to something that actually ships, and carrying it from "it ran" to "someone can rely on it."
The single completed project is the tell. It was not the most ambitious prompt. It was a bounded one, $339, a search database with a clear shape, and it crossed the line because the shape was small enough to finish.
This is the same wall Lovable's numbers exposed this week. $400M in revenue, 95% from individuals generating their own software, and a recurring complaint about paying for the AI's own failed fix attempts. Prototypes are cheap. Maintainable systems are where the value sits. The gap between them is the whole job.
FablePool is that gap printed on a public ledger. The prompt is a dollar. Everything that makes it worth more than a dollar is still you.
Sources: fablepool.com (project ledger, funding mechanic); Hacker News thread 48496539 (311 points, 174 comments, top comments). Pulled 2026-06-11.