Your daily English tutor in Telegram.
LinguaMate is a personalized AI English learning agent for Indonesian speakers that teaches through daily interactive lessons, progress tracking, and adaptive review. It starts with a placement assessment, builds a custom learning roadmap, and proactively nudges the user to practice in short sessions that fit real life.
Rina, Indonesian office worker, 29 - Rina has basic English exposure from school but freezes when writing messages or speaking at work. She wants small, practical lessons she can finish after work without feeling overwhelmed.
Dimas, tech-savvy student, 22 - Dimas uses Telegram often and wants to improve English for gaming, tech content, and future job opportunities. He prefers interactive practice over passive reading.
Ayu, returning learner, 35 - Ayu learned some English years ago but has become inconsistent. She needs reminders, measurable progress, and a system that helps her restart without embarrassment.
Rina opens LinguaMate on Telegram after work. The bot asks a few placement questions, learns that she struggles most with past tense and practical vocabulary, then creates a daily plan that fits 20 minutes before dinner.
The next day, LinguaMate sends a short lesson on useful work vocabulary and a simple sentence-building exercise. When Rina makes a tense mistake, the bot explains it in plain English and Indonesian, stores the error, and brings it back later in review instead of assuming she has mastered it.
After two weeks, Rina can see her streak, weekly accuracy, and recurring mistakes. She is still studying in short sessions, but now the product is helping her build consistency and measurable improvement, which increases retention and makes the learner likely to stay engaged long term.
Team & resourcing - Small team - 2 engineers, 1 product designer, part-time PM, and optional QA support.
Paste this into Cursor, Bolt, Lovable, or v0 to start building.
Build a personalized AI English learning agent called LinguaMate. Product summary: A Telegram-first English tutor for Indonesian speakers that proactively sends daily 15 to 30 minute lessons, starts with a placement assessment, adapts to the learner’s level, tracks vocabulary and grammar progress, and uses spaced repetition plus conversational practice to build long-term consistency. Future expansion should support WhatsApp, voice messages, and a web dashboard, but MVP must be Telegram only. Core MVP features: Telegram bot onboarding with /start, placement assessment, preferred study time, timezone, and daily duration Personalized daily lesson generation using Claude API Vocabulary lessons with Indonesian translation, part of speech, pronunciation guide, example sentences, and exercises Grammar lessons with explanation, examples, mistakes, and correction feedback Conversation practice modes with adaptive follow-up questions Basic spaced repetition for vocabulary review and recurring mistake review Progress tracking for streaks, lesson completion, vocab reviewed, accuracy, and weekly summaries Proactive reminder scheduler with pause/resume/change time controls Admin-friendly logging and basic usage analytics Suggested architecture: Next.js and TypeScript backend Telegram Bot API webhook integration Claude API for tutoring and structured outputs Supabase Postgres for persistent storage Redis plus BullMQ or a similar job queue for scheduled reminders and lesson jobs Schema validation for all AI outputs Row-level security and secure secret management Primary screens/flows: Telegram start and onboarding flow Placement assessment flow Daily lesson flow with quick reply buttons Review flow for spaced repetition Conversation practice flow Weekly progress summary flow Settings flow for reminder time, duration, pause/resume, and timezone Data model: User profile with Telegram ID, timezone, preferred time, daily duration, target level, interests, and status Assessment results with skill breakdown and baseline level Lesson sessions with date, type, generated content, completion status, and AI metadata Vocabulary items with translation, examples, due dates, confidence, and review history Grammar mistakes with error category, corrected form, and recurrence count Progress events and weekly summary records Reminder schedule records and delivery logs Implementation requirements: Use structured JSON outputs from the AI and validate them before sending to users Keep messages concise, mobile-friendly, and interactive Support bilingual English plus Indonesian explanations where helpful Make the learning engine platform-agnostic so WhatsApp can be added later without rewriting core logic Include basic error handling for AI failures, reminder failures, missing profile data, and invalid user replies Build the MVP so it is production-minded, maintainable, and ready for iterative expansion.
I want to build a personalized AI English Learning Agent that operates through Telegram initially, with WhatsApp support as a future expansion. The main purpose of this product is to help me, an Indonesian speaker who wants to improve my English gradually, consistently, and practically through daily interactive learning. This should not be a basic chatbot that only responds when asked questions. It should function as a proactive personal English tutor that guides me every day, assigns learning tasks, tracks my progress, remembers my weaknesses, and adapts the learning experience to my actual ability. PRODUCT VISION Build an AI-powered personal English tutor that acts like a patient, supportive, and intelligent English teacher. The agent should help me develop: - Vocabulary - Grammar - Reading comprehension - Listening comprehension - Writing - Speaking - Conversation skills - Sentence construction - Confidence in using English - Long-term learning consistency I want to improve my English from my current level toward intermediate and eventually advanced proficiency. The learning experience must be gradual, practical, interactive, and personalized rather than overwhelming or based on memorizing vocabulary without context. TARGET USER The initial target user is me, an Indonesian speaker who may struggle with English vocabulary, grammar, sentence construction, comprehension, and speaking confidence. The system should begin with an English placement assessment to estimate my current ability. It should identify my strengths and weaknesses and use the results to create a personalized learning roadmap. The agent must not assume that I am a complete beginner or an advanced English speaker without assessing me first. DAILY LEARNING EXPERIENCE Every day, the AI agent should proactively send me a personalized English learning plan through Telegram. I want the agent to tell me: - What I need to learn today. - Why I need to learn it. - What exercises I should complete. - How much time I should spend learning. - What mistakes I need to review. - What progress I have made. The default daily learning duration should be approximately 15–30 minutes, but I should be able to customize it. The AI should send daily reminders at my preferred time, respect my time zone, and allow me to pause or change my learning schedule. Each daily lesson should be interactive instead of sending a large block of information without requiring participation. Potential daily activities include: - New vocabulary. - Grammar lessons. - Sentence-building exercises. - Reading practice. - Listening practice. - Conversation practice. - Review of previously learned material. - A short daily assessment. The AI should adapt the number and difficulty of activities to my learning ability and available time. VOCABULARY LEARNING Every day, the agent should teach me useful English vocabulary with practical examples. The system should introduce approximately 5–10 new words per day by default, with the ability to adjust this based on my performance and retention. Each vocabulary lesson should include: - English word. - Indonesian translation. - Part of speech. - Pronunciation guide. - Example sentences. - Indonesian translation of example sentences. - Common expressions and collocations where relevant. - Similar or related words when useful. - Interactive exercises. For example: Word: Improve Meaning: Meningkatkan / memperbaiki Part of speech: Verb Example: I want to improve my English every day. Translation: Saya ingin meningkatkan bahasa Inggris saya setiap hari. Practice: Ask me to create my own sentence using the word. The agent should prioritize vocabulary that is useful in real-life conversations, work, technology, gaming, shopping, travel, and other practical situations. It should also explain differences between similar words, formal and informal expressions, and common mistakes. VOCABULARY RETENTION AND SPACED REPETITION The agent must remember vocabulary that I have learned and review it regularly. Implement a spaced repetition system that adapts to my performance. The system should track: - Words I have learned. - Words I frequently forget. - Correct and incorrect answers. - Review history. - Vocabulary confidence. - Words that require additional practice. Previously learned vocabulary should appear in future exercises and conversations. The agent should not consider a word mastered merely because I answered it correctly once. GRAMMAR LEARNING The agent should teach English grammar progressively, starting with the fundamentals and increasing difficulty based on my demonstrated understanding. Potential topics include: - Basic sentence structure. - Subject and object. - Verb "to be". - Simple Present. - Present Continuous. - Simple Past. - Future forms. - Questions and negatives. - Articles. - Prepositions. - Modal verbs. - Present Perfect. - Conditional sentences. - More advanced sentence structures. Every grammar lesson should contain: 1. A simple explanation. 2. Practical examples. 3. Indonesian explanations when necessary. 4. Common mistakes. 5. Interactive exercises. 6. Sentence creation. 7. Personalized feedback. When I make a mistake, the AI should explain what is wrong, why it is wrong, and how to correct it. It should distinguish between: - Grammatically incorrect sentences. - Grammatically correct but unnatural sentences. - Informal and formal expressions. - Acceptable variations in English. The teaching style should be patient and encouraging. CONVERSATION PRACTICE The AI should help me practice English through interactive conversations. It should ask questions, wait for my responses, and continue the conversation naturally. Conversation difficulty must adapt to my current ability. Potential conversation topics include: - Daily routines. - Hobbies. - Gaming. - Technology. - Work. - Shopping. - Travel. - Social situations. - Opinions. - Storytelling. - Real-world roleplay. Include different practice modes: - Casual English conversation. - Beginner conversation practice. - Grammar-focused conversation. - Vocabulary-focused conversation. - Roleplay scenarios. - Speaking confidence practice. - Job interview practice. The AI should encourage me to respond in English. It should correct important mistakes without interrupting every sentence unnecessarily. After a conversation, it should provide a brief summary of my performance, recurring mistakes, and recommended practice areas. LISTENING AND SPEAKING If Telegram voice messages and the selected AI services support these capabilities, allow me to send voice messages and receive voice-based English learning exercises. Potential features: - Speech transcription. - Conversation through voice messages. - Listening comprehension exercises. - Audio-based vocabulary practice. - Feedback on understandable pronunciation patterns, where supported by the available technology. The system must clearly distinguish between speech transcription and reliable pronunciation assessment. It should not claim to provide accurate pronunciation scoring if the technology cannot support it. PROGRESS TRACKING The agent should maintain a persistent learning profile and track my progress over time. Track: - Learning days. - Learning streak. - Completed lessons. - Vocabulary introduced. - Vocabulary reviewed. - Vocabulary retention. - Grammar exercise accuracy. - Reading performance. - Listening performance when supported. - Conversation practice frequency. - Repeated mistakes. - Areas that need improvement. The agent should provide weekly progress summaries and explain what I should focus on next. I want to see measurable improvement rather than simply receiving motivational messages. The system should use periodic assessments to evaluate changes in my English skills. It must distinguish between learning activity, exercise performance, and demonstrated proficiency. PERSONALIZED LEARNING The agent should personalize lessons based on: - My current English level. - My learning goals. - My available daily study time. - My interests. - My previous mistakes. - My performance. - My learning consistency. The AI should remember my learning context across sessions. For example, if I repeatedly make mistakes with Simple Past, the agent should provide additional explanations and exercises instead of continuously introducing unrelated topics. The learning plan should evolve as I improve. PROACTIVE DAILY NOTIFICATIONS The Telegram agent should proactively send daily learning notifications at my preferred time. I should be able to: - Set my preferred learning time. - Change the daily study duration. - Pause reminders. - Resume reminders. - View today's lesson. - Review previous lessons. - Request additional practice. The system should avoid sending excessive or unwanted notifications and should respect my notification preferences. TECHNOLOGY PREFERENCES I prefer the following technology stack for the product: - Telegram Bot API as the initial messaging platform. - WhatsApp Business Cloud API as a future integration. - Claude API as the primary AI model. - Next.js and TypeScript for the backend and optional web dashboard. - Supabase PostgreSQL for persistent user data, learning progress, vocabulary, and review schedules. - A reliable background job and scheduling system for daily reminders. - Structured AI outputs and schema validation where practical. The architecture should separate the learning engine from the messaging platform so WhatsApp can be added later without rewriting the core application. TECHNICAL EXPECTATIONS Design the product as a reliable, maintainable, and cost-conscious AI learning application. The system should support: - Persistent user profiles. - Conversation context. - Learning progress storage. - Vocabulary review scheduling. - Daily lesson generation. - Error handling. - Scheduled notifications. - AI usage and cost monitoring. - Secure storage of user data. Do not assume that all integrations are already configured. Identify required external services, API credentials, platform limitations, and implementation dependencies. MVP PRIORITIES Prioritize a functional Telegram MVP before advanced features. The initial MVP should focus on: 1. Telegram bot integration. 2. User onboarding. 3. English level assessment. 4. Personalized daily learning plans. 5. Vocabulary lessons. 6. Grammar exercises. 7. Interactive conversation. 8. Basic spaced repetition. 9. Progress tracking. 10. Scheduled reminders. Future features may include: - WhatsApp integration. - Voice-based learning. - Pronunciation feedback. - Web dashboard. - Gamification. - Advanced analytics. - More sophisticated adaptive learning. SUCCESS CRITERIA The product should help me build a consistent daily English learning habit and demonstrate measurable improvement in my language skills. Focus on: - Daily and weekly learning consistency. - Lesson completion. - Vocabulary retention. - Grammar accuracy. - Conversation practice. - Improvement in periodic assessments. - User satisfaction. - AI response quality. - Operating cost per active learner. Avoid vague claims such as "become fluent quickly." Define measurable learning outcomes and realistic assessment methods. PRD REQUIREMENTS Generate a comprehensive, practical, and implementation-ready PRD for this product. Include: - Product overview. - Problem statement. - Target user. - Product goals. - MVP scope. - User stories. - Functional requirements. - User journeys. - Telegram interaction flows. - AI agent architecture. - Database requirements. - Technical architecture. - Scheduling requirements. - Edge cases. - Security considerations. - Success metrics. - Development milestones. - Future expansion opportunities. Identify important assumptions and unresolved decisions. Prioritize a realistic MVP that can be developed and tested before expanding into a complex AI tutoring platform. The final PRD should be specific enough for an AI coding agent or developer to use as a foundation for implementation.
Design by The Resonance | Powered by GPC – The AI Transformation Company