Transparent product activity
Parlai activity evidence and counting methodology
Current, privacy-safe aggregate activity from Parlai with UTC definitions, source tables, limitations, and an explicit boundary against efficacy claims.
Current aggregate snapshot
As of July 14, 2026 at 8:02 PM UTC.
| Metric | Current value | Definition |
|---|---|---|
| Learner accounts | 9,353 | Accounts not scheduled for deletion |
| Active learners today | 66 | Accounts with an inbound message since 00:00 UTC |
| Messages exchanged today | 2,356 | User and assistant message rows since 00:00 UTC |
| Words learned this week | 279 | New or practiced vocabulary marked not previously known and not user-requested since Sunday 00:00 UTC |
Methodology
Counts are computed directly from Parlai's production users, messages, and vocabulary tables, cached for up to one hour, and rendered only as global aggregates. No phone numbers, names, conversation text, or country-level segments are exposed.
What these figures do not prove
These are product-activity measures, not an efficacy study. They do not demonstrate fluency, proficiency improvement, retention lift, or a causal learning effect. Account totals may include internal or test activity that cannot be reliably classified from the current schema. Recent UTC windows are partial until the period ends.