King2coding/ghana-ai-study-explainer

Affordable AI study explainer for Ghanaian learners, starting with JHS/BECE concepts in simple English and Asante Twi support. # Ghana AI Study Explainer An early web prototype for a curriculum-grounded Local-Language Study Explainer for Ghanaian learners. The first product wedge is: > Explain JHS/upper-primary school concepts in simple Ghanaian English, with draft Asante Twi support, Ghanaian examples, summaries, quizzes, and checks for understanding. ## Current Status This is a prototype, not a finished product. It currently includes: - A responsive web interface. - Five modes: explain, translate, summarize, quiz, check understanding. - Safety boundaries for leaked exams and high-risk advice. - A Node backend with provider support for local testing. - Free/near-free model paths: Gemini free tier, OpenRouter free models, local Ollama. - Paid/credit path: OpenAI API. - Template fallback when no provider key is configured. ## Run Locally ```bash node prototype/server.js ``` Then open: ```text 127.0.0.1 ``` ## Configure AI Provider Copy `.env.example` to `.env` and choose a provider. For Gemini free-tier testing: ```bash AI_PROVIDER=gemini GEMINI_API_KEY=your_key_here GEMINI_MODEL=gemini-2.5-flash-lite ``` For OpenRouter free-model testing: ```bash AI_PROVIDER=openrouter OPENROUTER_API_KEY=your_key_here OPENROUTER_MODEL=meta-llama/llama-3.3-8b-instruct:free ``` For local Ollama: ```bash AI_PROVIDER=ollama OLLAMA_MODEL=llama3.2 ``` For OpenAI: ```bash AI_PROVIDER=openai OPENAI_API_KEY=your_key_here OPENAI_MODEL=gpt-5.4-mini ``` ## Professional Roadmap 1. Add real source-pack retrieval from NaCCA/GES/WAEC materials. 2. Add feedback persistence and reviewer dashboards. 3. Add native-speaker Twi review workflow. 4. Deploy a private web pilot. 5. Test with 10-20 students, teachers, parents, and reviewers. 6. Update cost-benefit and investor documentation with real usage data.