Pragyanta Tutor – Grounded AI Learning with RAG

The description states that Pragyanta Tutor is a source-bound adaptive tutor for teachers, built using a Retrieval-Augmented Generation (RAG) approach. It allows teachers to upload lesson content (PDF or text), which becomes the system's source of truth.

• Students ask questions, and answers are generated based only on this approved material.
• Off-topic queries are declined instead of being guessed.
• The system supports a structured learning loop: Detect → Remediate → Verify.
• It includes:
  • Practice mode with multiple-choice questions
  • Teacher reports showing misconceptions and resolution status
  • A deterministic keyless mode for consistent evaluation without API keys

The system is described as running on:

• Frontend: React 18 + Vite
• Backend: FastAPI (Python)
• Database: PostgreSQL 16 with pgvector
• AI Layer: Pluggable provider (deterministic mock and optional live model)
• Infrastructure: Docker and Docker Compose

AI-assisted development tools used include OpenAI Codex and ChatGPT (GPT-5.6 series models).

Inference The system is a prototype built for demonstration purposes, likely intended to showcase how grounded AI can be applied in an educational setting.

The author claims that most AI tutors today answer from anywhere on the internet, which leads to:

• Unverified or off-syllabus answers
• Incorrect but confident responses
• Lack of visibility into student struggles

Pragyanta Tutor aims to address these issues by behaving like a real classroom assistant — one that teaches only from approved material and helps teachers understand learning gaps.

The positioning is:

• Controlled AI in education
• Evidence-based tutoring
• Trustworthy, measurable AI

It positions itself as an alternative to open-ended chatbots, emphasizing correctness over capability.

Inference This is a self-reported strategic shift toward trustworthiness and alignment in AI education tools. No external validation or market positioning data is available.

The description states that Pragyanta Tutor targets teachers who want to:

• Provide students with accurate, curriculum-aligned explanations
• Gain insights into student misconceptions and learning progress
• Use a system that ensures AI responses are grounded in trusted lesson content

It also mentions:

• A teacher report showing misconceptions, resolution status, and supporting evidence
• Support for multiple-choice practice questions derived from lesson content

Inference The primary ICP appears to be educators or instructional designers working within structured curricula, particularly those seeking tools that maintain alignment with their teaching materials.

There is no mention of pricing, monetization, or business model in the description. The project is presented as a hackathon submission and includes no indication of:

• Revenue streams
• Customer acquisition plans
• Subscription models
• Licensing or usage fees

Not evidenced

The system uses:

• Retrieval-Augmented Generation (RAG) to ensure grounded responses
• pgvector for vector storage and similarity search in PostgreSQL
• FastAPI + React for backend and frontend architecture
• Docker/Docker Compose for containerized deployment
• OpenAI Codex and ChatGPT (GPT-5.6) for development assistance

Key technical features include:

• Strict retrieval constraints to prevent off-topic answers
• Citation-style references for traceability
• Deterministic keyless mode for consistent evaluation
• Persistent learning state across detect → remediate → verify cycles

Inference The system is built with a focus on reliability, traceability, and reproducibility. It shows technical maturity in implementing RAG and maintaining structured interaction.

There is no evidence of:

• Real users or customers
• Revenue or ARR
• Product adoption metrics
• Customer feedback or usage data
• Production deployment history

The project is described as a hackathon submission, with a demo running in deterministic keyless mode for consistent evaluation.

Not evidenced

The description does not reference any competitors. However, it implies that current AI tutoring systems lack:

• Source-bound responses
• Traceability of answers
• Teacher visibility into student learning gaps

This suggests a gap in the market for controlled, curriculum-aligned AI tutors, which may overlap with:

• Existing EdTech platforms using AI
• LMS integrations with AI features
• Classroom assistant tools

Inference The project identifies a niche where correctness and alignment are prioritized over open-ended knowledge. No competitive analysis or market positioning data is provided.

1. No real-world usage or traction: The system exists only as a prototype, with no evidence of actual classroom use.
2. Single-person team: Limited development capacity and scalability concerns.
3. Hackathon origin: Likely not production-ready or designed for enterprise-scale deployment.
4. Deterministic keyless mode: While useful for evaluation, it may not reflect full functionality in live environments.
5. No pricing or monetization strategy: Unclear path to commercial viability.
6. Unverified claims: All assertions are self-reported and uncorroborated.

1. Has the system been tested with actual teachers or students?
2. What is the expected timeline for moving from prototype to production-ready tool?
3. Are there any plans to integrate with existing LMS or curriculum platforms?
4. How does the system handle different formats of lesson content (e.g., videos, interactive media)?
5. What are the long-term goals for scaling beyond a single classroom or subject?
6. Is there any plan to incorporate live AI models alongside the deterministic default?