Gizmo

Spaced-repetition as a product core but likely operationalized as a data-driven SRS: the marketing claim that 'days make your time twice as effective' implies they tuned spacing curves aggressively using empirical outcome data (not just SuperMemo/Anki defaults). This suggests they run experiments and fit per-item/per-user retention models (e.g., parametric forgetting curves or survival models) rather than one-size-fits-all intervals.

Rapid calibration / cold-start optimization: claims of very fast grade improvement point to a short, high-signal onboarding that quickly estimates user ability and item difficulty — likely via brief diagnostic quizzes, Bayesian skill estimates, or adaptive testing — enabling meaningful personalization within days instead of weeks.

Curriculum alignment and outcome feedback loop: the product seems tightly aligned to exam systems (GCSE, AP, A-level). That requires heavyweight mapping between deck content and syllabus standards, plus closed-loop labeling of efficacy (linking app usage to reported grades) to optimize which decks and item types actually drive exam performance.

Automated content ingestion and canonicalization pipeline: to scale across subjects and curricula they almost certainly have tooling to convert source materials (notes, PDFs, teacher uploads, user-submitted Q&A) into normalized flashcard units. This implies NLP/NER pipelines, templated Q/A extraction, and deduplication heuristics to keep decks concise and curriculum-focused.

Multi-signal quality control and ranking: with large user-generated repetition sets, they likely combine crowd signals (upvotes, usage frequency), performance signals (item success rates, relearning rates), and ML quality estimators to surface the best decks and to retire noisy/ineffective items automatically.