The Research Ledger

Pre-register and run a real study of the production effect — does speaking an answer aloud during recall beat recalling it silently? — inside Sapio's own read-aloud loop, with consenting users as participants.

Doable now, solo: doing and publishing this needs no accreditation and no diploma — preprint servers (OSF / PsyArXiv) have zero gate. The only real requirement is a genuine, pre-registered protocol.

Post the preprint and the open, anonymized dataset, then wire the outcome into this public ledger. "Money in, science out" becomes verifiable rather than a promise.

Transparency is the whole pitch — the result ships whether it confirms the effect or not.

With explicit informed consent, embed small randomized content-formulation contrasts into everyday review — and capture the signal almost no one has: did a real person actually remember, two weeks later?

Outcome-labeled retention at scale is the unique asset — most edtech optimizes proxies (clicks, time-on-app); this optimizes whether learning actually stuck.

Pre-registered replications of the other effects Sapio relies on — pretesting, spacing, retrieval practice, recall-by-analogy — each run in the open and wired into the public ledger.

Once the funding supports it: register a research vehicle (a fonds de dotation), bring on a credentialed lead researcher, and use the French R&D framework (the CIR tax credit; the agrément CIR to route third-party funding) so the loop is state-recognized, not just self-declared.

None of this is required to start — it is the step that makes the funding loop tax-real and audit-proof once money is flowing.