Tcher – AI-Powered Learning

Tcher helps every student learn their way, with personalized AI support that adapts to what they know, where they struggle, and even brings quality learning to rural communities

Comment

We started Tcher because we noticed that students don't all learn in the same way. One student might understand a concept immediately, while another may need a different explanation or more practice. Teachers also have limited time to give every student completely personalized attention.

This problem is even more important for students in rural and underserved communities, where access to quality tutoring and personalized learning support can be limited.

Tcher is our attempt to solve this. It is an AI-powered learning platform designed to work like a personal tutor for each student. Instead of simply answering questions, Tcher looks at what a student knows, where they are struggling, and adapts the learning experience around them.

Students can interact with Tcher through conversations, voice, quizzes, and educational content. We are also building support for multiple Indian languages so that personalized AI learning isn't limited to students who are comfortable learning in English.

Google Gemini is at the heart of our AI experience. We use Gemini to understand educational questions and content, reason about the context of a student's learning, and generate personalized explanations and learning experiences. Its multimodal capabilities also help us work with different types of educational content rather than relying only on plain text.

We built Tcher using Python and FastAPI for the backend, with Streamlit for the interface. Around Gemini, we built our own learning layer using concepts such as a knowledge graph and student knowledge map to connect concepts, understand learning gaps, and determine what a student may need to learn next.

Some of the main things we have built include personalized AI tutoring, student knowledge mapping, adaptive quizzes, voice-based learning, multilingual support, OCR for educational content, spaced repetition, adaptive study planning, and teacher content sharing.

One of the biggest things we learned while building Tcher was that an educational AI system shouldn't just focus on giving the right answer. The more important question is what the student needs to learn next.

That changed how we approached the product. Rather than building another chatbot around an AI model, we wanted to build a learning system around the student, with Gemini providing the intelligence behind many of those interactions.

We also faced challenges around AI accuracy, multilingual interactions, maintaining context across learning sessions, and making the experience simple enough for both students and teachers. Working with Gemini helped us experiment with these challenges while building an experience that feels more like learning with a tutor than interacting with a generic chatbot.

Our long-term goal is to make personalized learning support available to every student, regardless of where they live or what resources they have access to. We especially want Tcher to be useful for students in rural and underserved communities who may not have access to personalized tutoring.

We believe a student's location should not decide the quality of education they receive. Tcher is our step toward making personalized learning more accessible to everyone, with Gemini helping us turn that vision into a real learning experience.

Built With

Share this project:

Updates