Inspiration
What it does How we built it FlowNotes came from noticing how often students get stuck in the middle of learning, not after. General AI tools can answer questions, but they’re not built to guide someone while they’re working through problems, reviewing slides, or trying to understand a concept in real time. I wanted a tool that acts like a study co-pilot—an overlay that quietly helps you stay focused, explains things instantly, and organizes your learning flow without switching tabs or losing attention.
How I Built the Project
I built FlowNotes as a Python desktop application with a floating, toggleable overlay. It integrates AI into the student’s existing workflow while staying out of the way.
Key components:
Smart On-Screen Overlay The app creates a small panel that sits on top of the user’s screen—no switching windows. It provides:
instant explanations for highlighted text
step-by-step walkthroughs of problems
quick definitions
simplified or advanced explanations depending on user preference
Context-Aware AI Support Users select a subject (math, physics, CS, etc.). FlowNotes adapts responses to that subject automatically—definitions, examples, explanations, diagrams.
Study Flow Tools The app can generate:
flashcards based on what the user is reading
quick quizzes
concept breakdowns
hazard markers (identifying parts the user struggles with repeatedly)
Lightweight Architecture The app is built in Python using:
a desktop UI framework
a transparent overlay layer
streamed AI completions for instant interactions
What I Learned
Building FlowNotes taught me a lot about design and user needs:
Minimizing friction is everything — students avoid tools that interrupt them.
Contextual prompting dramatically improves how helpful an AI feels.
UI simplicity beats feature complexity — the overlay had to feel invisible until needed.
Real-time interaction loops require careful prompt engineering to avoid slow or cluttered output.
Challenges I Faced
Building a non-annoying overlay Getting transparency, drag behavior, and layout right took many iterations.
Avoiding information overload Early versions spammed too much text. I learned to implement:
tiered explanations
condensed summaries
user-controlled detail levels (simple → intermediate → deep)
Standing out from general AI apps Since ChatGPT already answers questions, I had to make FlowNotes about workflow, not just answers. The overlay + real-time study support is what makes it unique.
Python UI limitations Some frameworks made the overlay laggy or glitchy. Picking the right tools and optimizing responsiveness was a big part of the challenge.
Log in or sign up for Devpost to join the conversation.