AI conversation practice — for classrooms and research

Two distinct AI personalities in one platform. Coach mode corrects mistakes in real time and explains the rule. Partner mode practices naturally as a peer — no interruptions, just conversation.

• •Coach: real-time error correction with explanations (grammar, vocabulary, register, pragmatics)
• •Partner: immersive peer practice — the AI stays in character throughout
• •Mode assignable per study cohort for controlled research designs
• •Fluency score computed from speech recognition across sessions

Every activity has a real-world context — job interview, doctor's visit, market shopping — with custom roles for both the learner and the AI. Instructors define goals; the AI evaluates them after every session.

• •Custom scenario context, AI role, and learner role per activity
• •Instructor-defined session goals evaluated automatically by the AI
• •CEFR-aligned proficiency estimate produced at every session end
• •Score delta, goal completion rate, and error rates stored to the learner record

Build a full curriculum, invite students, and watch their CEFR progress in one place. No spreadsheets. No manual tracking.

• •Courses organised by target language and CEFR level (A1–C2)
• •Invite students by email — accounts provisioned automatically
• •Per-student view: session history, score trends, goal achievement
• •Enable or disable Vocabulary, Gamification, and Research modules per organisation

The AI extracts vocabulary automatically from every session and saves it to the learner's personal word list. A built-in gamification layer — streaks, XP, and badges — keeps learners coming back.

• •Automatic vocabulary extraction: terms saved with definition and source session
• •Learners organise words into custom lists and mark favourites
• •Daily streaks, XP per session, level-up system, and milestone badges
• •Smart push notifications nudge inactive learners automatically

Run language acquisition studies with the rigour your ethics committee expects — from enrollment to structured export, the platform tracks everything.

• •Cohort assignment: Coach-only, Partner-only, or randomised
• •Dose tracking: valid session count against a configurable target
• •Cohort comparison: score delta, fluency, goal completion, error rates
• •Week-by-week timeline chart of session volume and cumulative enrollment

Design pre/post assessments with 5 question types and automatic scoring. Export everything as CSV or JSON. Ask your data questions in plain language using Perplexity or GPT.

• •5 question types: Likert, Single/Multiple Choice, Open Text, Numeric
• •Automatic 0–100 normalization, reverse scoring, and multi-subscale support
• •CSV/JSON export: participant summary, session detail, questionnaire responses — with anonymization
• •AI research queries powered by Perplexity Sonar (live web search + citations) or OpenAI

Run as many isolated organisations as you need — each with its own users, courses, settings, and data. Role hierarchy keeps the right people in the right seats.

• •Four roles: System Admin, Institution Admin, Instructor, Student
• •Per-org feature flags, student caps, and access validity windows
• •Org-level behavioural analytics: DAU/WAU/MAU, completion rates, dropout signals
• •Org switcher so system admins can manage any tenant without re-logging in

You choose which AI powers each part of the platform. Route conversations to GPT-4, research queries to Perplexity Sonar, and translation to a cost-efficient model — all configurable per organisation.

• •Supports OpenAI, Anthropic, OpenRouter, Perplexity, and any OpenAI-compatible endpoint
• •Assign models by platform area: conversations, research queries, translation
• •Per-org override — your research institution can use a different model than your schools
• •Self-hostable: full stack on Docker Compose, no mandatory cloud dependency