Sahayak

Sahayak is a multilingual, AI-powered co-teacher built for rural India’s multi-grade, low-resource classrooms. It uses Gemini’s multimodal intelligence with an agentic architecture to generate hyper-local teaching content, differentiate worksheets from textbook photos, answer student/teacher questions in local languages, plan lessons, and run voice-based reading assessments—all with an offline-friendly, pedagogy-aware design.

Project Details

📌 Problem Statement: In many Indian government schools, a single teacher manages multiple grades and subjects in one room with limited resources. Existing LMS/classroom tools are not optimized for hyper-local content, multilingual delivery, or mixed-ability groups, creating gaps in engagement, differentiation, and speed of preparation.
💡 Solution: Sahayak deploys teacher- and student-facing AI agents that accept text, images, and voice. Teachers generate hyper-local content, differentiated worksheets, and weekly plans in minutes; students receive age-appropriate, language-matched explanations and practice. A reading agent conducts quick voice-based fluency checks to monitor progress.
🏗️ Architecture: Genkit orchestrates Gemini tools behind a modular, agentic interface. Teacher/Student agents invoke capabilities (content generation, worksheet builder, blackboard converter, Q&A, planner, reading assessor). Firebase handles auth, data, serverless logic, and hosting. Multimodal pipelines (Vision + STT) support real classrooms where inputs are photos and speech, not just typed text.
🧠 Pedagogy & UX: Outputs are analogy-led, level-aware, and multilingual to fit mixed-age groups. The blackboard integrates creation→assessment loops (sketch → quiz). Differentiation focuses on scaffolded difficulty, varied question types, and localized examples to enhance comprehension and retention.
🔌 Extensibility & Ops: Low-bandwidth/offline-fallback modes prioritize reliability in rural contexts. Integrations via n8n/A2A support automated class updates and future workflows (e.g., exporting plans, attendance nudges). Firebase Studio streamlines deployment and monitoring for rapid iteration.
🌍 Future Plans: Voice assistant mode for hands-free classroom control; math problem generator; deeper n8n agent automations; parent dashboards for progress visibility; broader language coverage and richer reading analytics.
🚀 Vision: Make high-quality, culturally relevant instruction universally available by turning any classroom—regardless of grade mix, bandwidth, or language—into a personalized, AI-supported learning environment.