This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend
The Notebook Nobody Can Read
Every student knows this moment.
The lecture is moving fast. The professor is three slides ahead, and you're writing as quickly as your hand allows. Words lose their vowels. Sentences lose their endings. Arrows point from one idea to another, and you promise yourself you'll remember what they mean.
Two hours later, you open your notebook to revise and stare at a page that might as well be written in another language.
This is our friend Surraya. She's a fourth-year psychology student, and this is her, every week, every course, every exam season.
The frustrating part is that the knowledge is there. She was in the lecture. She paid attention. She wrote it all down. But by the time she sits down to study, she's spending her energy decoding her own handwriting instead of learning the material, and she's already worn out before the real studying even starts.
So, to be a good friend, we built her an app.
We called it NOTE-IT.
Why a Generic Chatbot Wasn't the Answer
The obvious suggestion is "just paste your notes into an AI chatbot." We thought about that, and it didn't fit Surraya for three reasons.
First, lecture notes are personal. They're full of half-formed thoughts, questions, and things she doesn't understand yet. Uploading all of that to someone else's server every week isn't something she should have to do just to study.
Second, a chatbot gives you a wall of text back. Psychology notes have a shape: key terms, classic studies, researchers, real-life examples, and the processes that connect them. A generic summary flattens all of that, so we built NOTE-IT around how psychology is actually studied.
Third, students don't have API budgets. A tool that charges per request, or needs a subscription, isn't a tool a student will keep using.
We wanted something private, free, and shaped around how studying actually works. That meant running an open model locally, on her own laptop.
How NOTE-IT Works
Surraya pastes in her notes exactly as she wrote them, typos and all, and hits Generate. Everything after that happens on her laptop:
Messy lecture notes (pasted in)
│
▼
┌───────────────────────────────────┐
│ Gemma 3 (4B), running locally │
│ through Ollama │
└───────────────┬───────────────────┘
│
Reads the notes and sorts them
into a psychology study template
│
┌───────────┼─────────────┬──────────────┐
▼ ▼ ▼ ▼
Organized Flash cards Diagram Study buddy
notes (flip to (Mermaid) (Tutor, Peer,
(template) test yourself) Examiner)
Gemma does every AI step. It reads the notes and rewrites them into a template built for psychology, with key terms, important studies, and real examples each in their place.
Then there's the part Surraya liked the most: a little robot study buddy. It has three modes:
- Tutor, for when she needs something explained
- Peer, for talking it through like a classmate
- Examiner, for when she wants to be tested
Whichever mode she picks, the buddy answers from her own organized notes, not from random stuff on the internet. We give Gemma her notes as context, so it stays grounded in what her professor actually taught.
There are also notebook tools for working with her notes, and a calendar so she can plan her revision before exam week sneaks up on her.
Demo
Paste messy notes, hit Generate, get organized notes in a subject template.
![Flash cards]
Flash cards built from her own notes. Click to flip.
The key process drawn as a diagram.
The robot study buddy, answering from her notes.
Putting It to the Test with Surraya
This weekend, we handed NOTE-IT to Surraya. We deleted every saved note so she was starting from scratch, opened Ollama, and started the app.
She pasted in her own psychology lecture notes and readings and hit Generate. The first run takes about 35 seconds, because Gemma is reading, sorting, and rebuilding everything right there on her laptop. We'd warned her about that, and she watched the robot while it worked.
Then her notes came back, organized, and she went straight to asking the study buddy questions about what she'd written.
When we asked what she thought, she said:
"I used it this weekend for my psychology lecture notes and readings and honestly I'm kinda impressed!!!!!!! It runs on my own computer, so my notes never leave it, and I can just ask questions about what I wrote without having to paste everything into some website. I also really like how it works.I ask a question and it actually finds what I mean. The interface is super clean and easy to use too. It's honestly so cool that my friends built this, and I'm really excited to see what they add next!!"
What Was Hard
The biggest trade-off was speed. Running a model locally means a full Generate takes about 35 seconds on a normal laptop. A cloud API would be faster. But a cloud API would also mean uploading Surraya's notes, needing an account, and paying per request. For a study tool a student uses every week, we decided a short wait was worth it if her notes never leave her computer. Hearing her say "my notes never leave it" told us that was the right call.
Code and Running It Yourself
4leena
/
NOTE-IT
Turns messy lecture notes into subject-specific study notes with diagrams and a study buddy chatbot. Runs locally with Gemma via Ollama.
NOTE-IT
Paste your rough lecture notes and get back notes you can study from, a diagram, and a study buddy that quizzes you. It all runs on your own laptop with Gemma and Ollama, so you can study offline, with no tabs or notifications pulling you away.
We made it for a friend who studies psychology, for the DEV Hacktoberfest Weekend Challenge ("Build for a Friend"), in the Best Use of Gemma category.
What it does
NOTE-IT sorts your notes into key terms, theories and researchers, key studies, evaluation, real-world examples and a short quiz. It also draws a diagram of how the ideas connect.
The Study Buddy answers your questions from your own notes. It has three styles. Tutor explains things step by step. Peer talks like a friend who took the same course. Examiner asks you a question, marks your answer and tells you what you missed.
If…
To run it locally:
# 1. Install Ollama from ollama.com and pull the model
ollama pull gemma3
# 2. Clone the repo
git clone https://github.com/4leena/NOTE-IT.git
cd NOTE-IT
# 3. Start the app
uv run streamlit run app.py
Then open the local address Streamlit shows in your browser.
Stack: Python, Streamlit, Ollama, Gemma 3 (4B), Mermaid
Credits: Built by Aleena Mukeer and Ahad Ali Baig (@notahead). The robot mascot was drawn for this project, inspired by stock robot illustrations. Diagrams use Mermaid (MIT), icons are from Lucide (ISC), and the fonts are under the SIL Open Font License. We used Claude Code to help connect the pieces and build parts of the code, and the demo voiceover was made with ElevenLabs.
Why Open Innovation Matters Here
NOTE-IT only works the way it does because Gemma is an open model.
Because it runs on Surraya's laptop, her notes stay hers. There's no account, no server, and no API bill. It works on bad campus Wi-Fi, in the library basement, or on the bus home with no internet at all.
And because the model and the code are open, NOTE-IT isn't finished when we stop working on it. Someone can swap in a different model, extend it to other subjects, or adapt it for their own friend who can't read their notes either.
Building for Someone You Know
Building for a friend changes how you make decisions. We didn't add features because they looked impressive in a list. Every choice came back to one question: will this help Surraya actually study, instead of spending her evening trying to read her own handwriting?
Her favourite part wasn't the feature we expected. It was being able to ask questions about her own notes and get answers that actually understood what she meant. That's where we want to take NOTE-IT next.
If you've ever opened your notebook and not recognized your own notes, we'd love to hear how you deal with it. Drop a comment below.
Top comments (0)