01 · a lab, not an agency
you do not need a spec, a vendor shortlist, or a technical co-founder. you need the thing described plainly & somebody who will build it inside a day.
02 · the follow-ups
every one of these changes the estimate. the last two change it a lot .. connectors are where the time actually goes.
03 · what happens next
plain words. the follow-ups do the shaping, not you.
the estimate comes off the scope signals. it will be well under what an agency quotes, because the work is assembled from connectors rather than written from scratch.
public. anyone can read it, vote it up, or argue with it.
competing claims are the feature. two teams building the same brief is how you find out which reading of the problem was right.
funded but not paid. nobody is trusting anybody yet.
code, connectors, scalability & security get checked before anything reaches you. then the choice is yours.
the team you picked takes 70 percent. the teams you did not still share in it. that is deliberate.
04 · the split
published, versioned, & enforced by the database rather than by a promise. the policy row that applied to a payout can be read back years later.
05 · where this could fail
it reads four signals & multiplies. a build that looks like two connectors & one screen can turn out to be one connector with an undocumented api & three weeks of guessing. the number is an opening position, not a price, & the factors are public so you can see exactly how little is behind it.
ten percent split equally across every team that entered is generous at four teams & meaningless at forty. it is a motivation mechanic & motivation mechanics decay. if this stops working the policy gets a new version — visibly, with the old one still readable.
the lab is open source & non-profit. the ten percent that is not paid out sits with the org to fund the next round of builds, which is the only reason it exists.