Why I’m Building Clarity AI

I was building a tool called "Syllabus AI." The whole idea was to let students chat with their syllabus PDF. And yeah, that’s cool. It solves a real problem. Nobody reads PDFs.

But then I thought about my own time in university. Was the syllabus really the biggest pain point? Or was it the fact that my entire education was scattered across 50 different files?

You have the syllabus in one tab. The lecture slides in another. Your messy notes in a Google Doc. A random quiz on the LMS. It’s digital chaos. You spend more time Command+Tab-ing than you do actually studying. So I decided to expand the scope of what I was already building. Because we don’t just need a "Syllabus Reader." We need a Course Brain.

What is Clarity AI? Imagine if you could take everything such as the syllabus, the slide decks, the readings, the quizzes, and dump them into one massive, intelligent bucket.

Then, you just ask one question:

"How does the professor's lecture on Neural Networks connect to the Chapter 6 reading?"

And instead of searching through three different PDFs, Clarity AI just... tells you. It connects the dots between the slides and the textbook. It’s like having a TA who has memorized every single document in the course.

Technical Approach & Public Build Clarity AI is being built as a modern, full stack SaaS application. I believe in building in public to gather feedback early and document the engineering challenges of applying AI to real world educational data.

The Future Roadmap Phase 1 is focused on unifying text and slide based content. However, the vision extends to making course data truly accessible across modalities:

• Audio Interfaces: Implementing text to speech to generate daily "briefings," allowing busy students to consume course updates and summaries like a podcast while commuting.
• Multimodal Analysis: Integrating vision models to deeply parse diagrams, charts, and visual data within lecture slides, making them as queryable as text.

Clarity AI is not designed to replace educators. It is designed to remove administrative friction and cognitive load, allowing students to focus on learning.

The Stack I’m building this in public, and I’m going with the "Speed Run" stack:

• Frontend: Next.js (because I need this UI to feel buttery smooth).
• Backend: FastAPI & Python (Python is non-negotiable for the AI stuff).
• The Magic: A custom RAG pipeline that I’m tuning to handle messy, real-world data.

This is going to be a fun build. I’m starting from zero. Let’s see how far we can push this.