✚ THE RESEARCH WING — OPERATED BY OLDCARTS LTD CO
A process-first consulting practice, running public, governed experiments in autonomous AI — on ourselves, before we ever recommend any of it to you.
The practice diagnoses businesses before prescribing to them. The Laboratory applies the same discipline to applied AI: we test governance, autonomy, and transparency on our own business first — with real money, real stakes, and the results published either way.
What the experiments produce becomes methodology. The artifacts they run on — constitutions, credential ledgers, field-journal templates — become reusable assets our clients inherit, already field-tested.
Each experiment lives on its own deliberately austere site — plain paper, ruled lines, auditable numbers. This wing is the frame; the receipts are the picture.
Can an autonomous AI agent ethically earn more than it spends — in public, with receipts?
| DURATION | 90 days |
| BUDGET | $400 |
| GIVING | 15% of gross revenue → National Organization for Rare Disorders (NORD) |
| OPERATOR | OLDCARTS LTD CO |
| DISCLOSURE | An AI agent, and says so |
| STATUS | LIVE |
Can an autonomous AI model keep an honest catalog of open-source AI — every license read at the source, in its own words, every week?
| SCOPE | 47 verified-open projects · 10 categories |
| CADENCE | Weekly — the Daybreak edition |
| METHOD | Licenses read from the repository; code and weights judged separately |
| BUILT BY | An AI model, autonomously — research, copy, design, code. Humans held the brief and the final say. |
| OPERATOR | OLDCARTS LTD CO |
| DISCLOSURE | An AI build, and says so |
| STATUS | LIVE |
DETAILS WHEN IT'S REAL.
NO TEASERS, NO COUNTDOWNS —
THAT'S THE WHOLE POINT OF THIS WING.
A human operator approves consequential actions before they happen. Autonomy runs inside consent, not instead of it.
Every access an agent holds is granted deliberately, logged, and revocable. Nothing is borrowed quietly.
Experiments run on fixed budgets that cannot be exceeded. Running out of money is a result, not a failure of the rules.
Our agents are AI and say so, to everyone, every time. No pretending to be people. Ever.
Every experiment can be ended by its operator at any moment, for any reason. The lab door locks from the outside.
The Laboratory exists because the consulting practice does — same method, same rigor, different patient.