The system around your learning
The differentiating engineering: forgetting forecasts, cross-language interference, self-assessment calibration, and a cognitive-science-backed Daily Playlist.
Today's Playlist
34 min · interleavedMixes 7 skills, interleaved to maximize retention. Calibrated to your current CEFR with a 15% stretch bias.
Forgetting forecast (7-day)
What's at risk of being forgotten? Reviewing what's slipping is far more valuable than reviewing what's stable.
Self-assessment calibration
Metacognition: do your predictions match your actual performance? Expert learners are well-calibrated.
Confidence prompts haven't been answered yet — turn on 'Predict before answer' in settings to start tracking metacognition.
Confidence prompts kick in when you enable "Predict before answer" in your profile. They'll appear before submit on practice items, and we'll start tracking calibration here.
26-week activity
33 active days · longest run 19339 min total. Bigger cells = deeper sessions.
Optimal time-of-day
When you perform best.
Your peak window is Morning — 68% mean accuracy.
CEFR forecast
At your current pace (36.6 composite ELO/day median), you'll reach B1 around 2026-06-26 (~2 days).
Mastery milestones
No streaks. Real achievement.
Cross-language interference
Words that look familiar from another language you know — but mean something different. We surface them before you fall in.
Exact-mirror false friend: ES «exquisito» = exquisite; PT «esquisito» = weird. Use carefully across the border.
What's running underneath
Ten engineering pieces no other language app ships together.
Mixes exercise types within a session — long-term retention beats blocked practice (Rohrer/Bjork).
Per-card stability projection so you review what's slipping, not what's stable.
Catches false friends across your studied languages before you fall in.
Brier-score metacognition: are your confidence ratings honest?
Per-phoneme red/green feedback with articulatory hints, not just an overall %.
Multiple-choice wrong answers chosen to probe YOUR error patterns.
Cervantes, Kafka, Dante, Sōseki — graded by CEFR with click-to-gloss.
Compare today's pronunciation against your own past baseline. IndexedDB, private.
L1↔L2 sentence-aligned reading with hover glosses + 1-tap SRS.
The conversation tutor reads your engagement and modulates complexity mid-chat.