Everything used to launch Backpay, public on purpose. If the data union makes sense to you, take any of this verbatim, remix it, post it — that's what it's for. Assets: the film · the live meter · the code.
Tweet 1 carries the film (attach the MP4 natively, not a link). Links go in the later tweets — X downranks link-first posts.
my AI coding tools logged $1,556.27 of usage in 65 days. I was paid $0 for that data. so were you. every tool ships your telemetry home and sells the story — to investors, to competitors, to anyone but you. today that ends. BACKPAY: a data union for AI dev tools. the meter is on.
how it works: → npx backpay → it reads the usage logs your tools ALREADY write to your disk → shows you every single byte before asking consent → counters only. no prompts, no paths, no repo names — the schema has no field for them one cron tick a day. no daemon. leave with one command.
the meter is open source — one file, zero dependencies, readable in one coffee. no aggregate covering fewer than 5 members is ever published or sold. individual rows: never, at any price. don't trust me. read the file.
the money: vendors, VCs and funds already pay for spend-side data (who paid). they're blind on usage (who stayed). backpay sells that index — and 50% of every contract flows back to members, pro-rata by days contributed. the ledger is public from day zero. right now it says €0.00. honestly.
I'm member №1 — my own 65 days of usage are the first rows in the data union, live on the homepage. the first 100 members carry permanent founding weight in the revenue share. backpay.me — claim your backpay.
news.ycombinator.com/submit → use the title verbatim (no exclamation marks, no marketing words — HN punishes them). Post the body as a first comment. Best window: weekday morning US Eastern.
Show HN: Backpay – a data union for AI coding-tool usage (I'm member #1)
Every AI coding tool writes usage logs locally, and most also ship telemetry home, where it becomes market intelligence that gets sold — without the dev ever being offered a cut. Backpay is my attempt to invert that: an open-source meter (one file, zero deps, builds on ccusage) reads your local logs, shows you the complete payload, and submits day-level counters only after explicit consent. The schema has no field for prompts, paths or repo names — it can't leak what it can't hold. The data union sells k-anonymous aggregates (k≥5 per cell) and 50% of every contract accrues to members pro-rata by days contributed, in a ledger you can query from day one. Default payout rail is Lightning. I'm member #1 — my own 65 days ($1,556 of estimated usage) are the first rows, live on the homepage. The ledger currently reads €0.00, because no contract has been sold yet; I'd rather show that honestly than launch with vapor numbers. Things I'd genuinely like this crowd's input on: panel-bias methodology (early members will skew heavy-usage), whether k≥5 is sufficient for cells this sparse, and the payout rail choice. Meter: https://github.com/Backpay-me Live data: https://backpay.me
The launch happens in the replies. Predicted objections, with answers — anyone is welcome to use these.
“spyware with better fonts.”
spyware hides what it takes. this prints every byte and asks first. the schema literally cannot hold content — no field exists for prompts or paths. one file of code, no daemon, leave in one command. the difference isn't the fonts, it's the architecture.
“ccusage already does this.”
ccusage is the reader — it shows YOU your own stats, and backpay builds on it (with credit, it's great). backpay is the economics layer: k-anonymous aggregation, buyers, and a ledger that pays you. ccusage tells you what you burned; backpay gets you paid for the telling.
“who actually buys this?”
the same buyers who already pay for spend-side data (Ramp's AI index proved the demand). spend tells you who paid. usage tells you who stayed. churn/retention at the tool level is the layer nobody can currently see — that's what's for sale. aggregates only, never rows.
“50% of zero is zero.”
correct — the homepage ledger says €0.00 today, on purpose. the data union exists before the first contract; that's exactly why the first 100 members carry permanent founding weight. you're early or you're not.
“why would I trust you?”
don't. the meter is one file of dependency-free javascript — read it in a coffee. run `npx backpay preview`: it prints everything that would be sent and sends nothing. trust here is structural, not reputational.
“what about Cursor?”
no clean local usage log — cursor buries token data in binary blobs. CLI agents first (claude code, codex, gemini, etc. all write readable logs). cursor lands when it can be done counters-only, or not at all.
“GDPR?”
consent-based (nothing sent before you type y), erasure built in (stop + email), full payload portability (export), k≥5 on everything published, plain-language policy: backpay.me/privacy.html — written before launch, not after the first complaint.
Shortest possible versions, for anyone who wants to spread it.
your AI tools sell your usage data and pay you $0. backpay is the data union that flips it: open-source meter, counters never content, 50% of every contract back to members. npx backpay → backpay.me
Every AI coding tool harvests usage telemetry and monetizes it — the developers generating that data were never offered a cut. Backpay is an open-source data union that inverts the model. A tiny meter reads the usage logs your tools already write locally, shows you every byte, and submits day-level counters only — no prompts, no paths, no code, by schema design. The data union sells k-anonymous aggregates (the live index of which AI tools are actually winning) and 50% of every contract flows back to members, pro-rata. Member №1 is the founder, dogfooding 65 days of his own data. The first 100 members carry permanent founding weight. backpay.me