October 2026
Launch readiness
- Streaming counterpart responses (first words in ~3 s)
- Safety protocol: deterministic detection of self-harm, harm-to-others and abuse disclosures (EN/中文), session lock, crisis resources, designated-contact notification
- Per-organisation monthly session allowance; per-user rate limits; CSRF protection
- Scenario language toggle (English / 简体中文) per learner
- Management (Enterprise) pack — 10 scenarios
- Full-library play-test (98 Practice scenarios) and counterpart calibration: first-turn restraint, utterance length cap, difficulty-based opening temperature
- Production watchdog with out-of-band alerting
- Early-access sign-in; Google Workspace SSO