Skip to content
Go back

Dev Log 0 - I feel it coming

Published:  at  12:06 AM

Hello, hi!

If you’re reading this, you’re probably my mom, or one of the first few curious folks keeping an eye on Nuclear. Either way, welcome. We’re an AI-powered note-taking app, and this is our very first dev log.

I’m Ojas, one of the four semi-responsible humans behind this project. We’ll be posting devlogs every week (ish) with our own twists, behind-the-scenes chaos, and team updates. You’ll get to meet the rest of the team soon; they’re awesome.


The idea

Nuclear started out like most of our “startup ideas”, somewhere around midnight, in a conversation we barely remember. Normally these ideas vanish by morning, but this one stuck. Maybe because we couldn’t stop complaining about all the existing AI tools. Maybe because we had too much ramen. Or maybe because we realized… we could actually build something better.


The problem

I’ve used almost every AI tool that claims to “boost productivity” for students. As someone studying Mathematics and Computer Science in university, I face two main problems:

  1. Long, brutal lectures that lose me halfway through.
  2. Even longer nights trying to catch up.

Naturally, I turned to tools like NotebookLM and KnowNotes.ai. But here’s the thing:

They’re not built for students. They’re built to look like they’re for students. And the more I used them, the more I heard the same frustrations from my friends. So we figured, “okay,what if we actually made something that understood how students learn?”


The premise

Every app out there follows the same formula:

Give us your input, and we’ll drown you in resources.

Summaries, flashcards, podcasts (??), even random questions, none of it helps if you’re just receiving information. Let’s be honest: if you wanted summaries, you could’ve Googled someone’s notes from 2008 and gotten something better than what these tools spit out.

The real problem?
These tools make learning passive. They spoon-feed you content and call it “productivity.” But real learning doesn’t happen when you’re being fed. It happens when you engage.

We knew we wanted to fix that. So we sat down, ranted for a few hours, and came up with a few rules.


The Solution

Don’t give students too much, too fast.

Imagine taking an exam with the answers already printed next to every question. Sounds easy, right? Now ask yourself—did you learn anything?

Our idea is to slow things down on purpose. If you want the AI to hand you a summary, first show us you actually understand the material. Fill in the blanks. Answer a few questions. Prove that you’re not just passively consuming.

Our mentor, who’s spent decades in education, backed this idea completely. He told us it mirrors how long-term memory works. It’s about recall, not recognition. And if we can turn this process into something rewarding? Even better.


Even the slop has to be good slop.

If AI is going to generate content, it better not be nonsense tied together with em dashes and wishful thinking. We’re building systems to vet, edit, and elevate what the LLMs give us; so summaries, quizzes, and flashcards are actually useful. That might mean adding constraints, post-processing rules, or even integrating multiple models.

This isn’t “Let the AI handle it.”
This is “Let the AI help and let us make sure it’s worth reading.”


Notetaking > AI

No matter how fancy the tech gets, the best way to learn will always be writing things in your own words. That’s why we’re designing a powerful note editor that supports:

…and a writing experience that actually feels good. Think of it as a cross between Notion, Google Docs, and Overleaf, but built from scratch, for students. You’ll be able to use your own notes as fuel for quizzes, flashcards, and more. Or don’t. They can just be your notes. That’s valid too.


Packing it all up

Everyone studies differently. We’ve identified three broad “modes” of learners:

  1. Give me the content, I’ll figure it out.
  2. Test me a little first, I want to earn it.
  3. Let me do everything myself. I just want a good editor.

We’re designing Nuclear to adapt to all three. You can toggle modes anytime, depending on your energy levels or the mood of your existential crisis. No judgment.


Design focus

One last thing: Nuclear is going to look good.
Notion changed how people feel about note-taking because it’s pretty. We believe in function and form. So we’re building every design from the ground up. No UI libraries. No generic templates.

Just handcrafted components, made by students, for students.
(Avan, our designer, would not let us sleep if we did otherwise.)


This week at Nuclear…

It’s been a week of planning. Long whiteboarding sessions. Diagrams. Feature breakdowns. Sitemaps. Midnight voice notes about dashboard structures. But we’re clearer than ever on what we’re building.

Next week, we start putting those plans into code.
We’ll be back soon with more updates (and probably bugs).

Thank you for being here.

May the force be with you —ojas


Suggest Changes

Previous Post
Dev Log 1 - The Under-Taker
Next Post
Userflow Diagram