
SwaVid
An AI tutor that learns your child before teaching them.
Screenshots


SwaVid
SwaVid is an AI-powered learning platform built for Indian school students, parents, teachers, and schools. The core premise: static content delivery doesn't account for where a student actually is in their understanding. SwaVid replaces it with adaptive, personalized instruction driven by diagnostics, AI orchestration, and real-time learner state.
The platform runs on a Next.js frontend, an Express.js backend, PostgreSQL, Redis, and a suite of AI providers including Google Gemini, Vertex AI, OpenRouter, and Sarvam for speech. A separate Python service handles live voice tutoring and game agent sessions.
A Cognitive Test profiles the student's learning style and working memory patterns. Chapter Diagnostics assess prerequisite knowledge before a chapter begins. Both feed a learning state model in PostgreSQL that personalizes everything downstream.
Lessons are adaptive: the backend loads the student's cognitive profile, checks cached content, queries a RAG service for context, and calls Gemini to generate a personalized explanation. A separate Next.js server route formats the result into whiteboard-style slides via Vertex AI. Quiz, revision, and practice systems draw on the same learner state to target misconceptions and adjust difficulty dynamically.
Learning Debt Identifier
The flagship public diagnostic, accessible without login at /learning-debt-identifier. The concept: a student in Class 9 may have unresolved gaps in Class 6 content that silently undermine their current learning. The LDI finds those gaps.
Curriculum Scan takes a student through up to 40 adaptive questions covering a full NCERT grade and subject, then produces a visual repair-map report: chapter heatmap, priority repair map, phased roadmap, evidence appendix. Questions are served from pre-composed quiz plans in PostgreSQL never from live AI calls during the session for consistency and speed. A single Gemini call optionally polishes the natural-language report layer.
Topic Readiness works at the chapter level: 10 questions targeting the prerequisite concept graph for one chapter, outputting a parent-facing readiness report with blocking/needs-recap/ready concept labels and suggested home activities. The catalog covers 220 topics across Classes 4–10 for both NCERT Science and Mathematics.
Thinking Coach
SwaVid's AI voice tutor. It uses a Socratic approach guiding students to reason through problems rather than giving direct answers. The live voice version runs as a separate Python process using the Vision Agents SDK, joining a Stream call and using Gemini realtime audio for speech-in, speech-out tutoring. The frontend has a typed adapter boundary that falls back to the REST-based Thinking Coach when live voice is unavailable.
Related products
View all
AntForms
Unlimited free submissions + free analytics + integrations +

AthleteMatrix
Your personal coach based on data from your everyday apps
More categories
All categoriesLaunched in Week 26, 2026
See the week
MailCull
Email verification that cleans your list and cuts bounces

Habit of the Day
One habit a day from our feed with no streaks!









