What if we could engineer the dopamine hit of endless scrolling into actually learning something?
Right now, recommendation algorithms just want to keep your eyeballs glued. When you try to learn a new tech concept, you hit a wall. You waste hours digging through tutorials trying to find one that matches your actual skill level.
I got tired of the friction, so my team and I built Learnora as a part of our university project. It’s an system that figures out your baseline and feeds you the right content at the right time.
Under the hood, it uses LLMs and web crawlers in a three-step loop.
First, a conversational agent maps your existing knowledge. It builds a custom learning path and explicitly filters out the concepts you already know.
Next, a discovery engine fetches and ranks multimodal content—videos, articles, podcasts—calibrated exactly to your current node on that path.
Finally, an evaluation layer tests your comprehension. Once it verifies you actually grasp the material, the system automatically advances you to the next node.
The project is completely open-source, and anyone is welcome to jump in, open a PR, and contribute to the codebase.
After all, we’ve already built the infrastructure to flawlessly hijack human attention. It’s time we stop using it to feed ourselves garbage, and start using it to actually make ourselves smarter.