Locus AI

Inspiration

Students spend hours reading notes, then realise they cannot recall or connect what they studied. We wanted to make progress visible and turn passive reading into active understanding.

What it does

Locus AI turns a PDF or DOCX into a growing 3D knowledge network. It identifies topics, subtopics, and relationships from the uploaded document. As students read, nodes unlock, source-linked flashcards appear, and short checkpoints test understanding. Mistakes shape later questions, helping students practise their weak areas.

How we built it

We built a document-to-graph pipeline that extracts text, identifies concepts, maps their relationships, and renders them as an interactive 3D network. The reading canvas stays connected to every node, so students can jump from a concept back to its exact source passage. We used AI for document-grounded topic analysis and adaptive question generation.

Challenges we ran into

The hard part was preventing the graph from becoming visual noise. We had to keep connections meaningful, make progress feel rewarding, and avoid interrupting someone while they were focused on reading. We also designed checkpoints to be optional and closeable.

Accomplishments that we're proud of

We created a study experience where the network grows only when learning happens. Each node is tied to the learner’s own document, not random outside content. The result is a personal map of what they have read, understood, and need to revisit.

What we learned

A beautiful graph alone does not teach. Students need clear source context, retrieval practice, and feedback at the right moment. The visual layer works best when it supports those actions.

What's next for Locus AI

Next, we will improve mistake-pattern detection, add spaced revision paths, support multiple documents, and build topic mastery insights. Our goal is to make every student’s study material feel less like a stack of pages and more like a place they can explore, test, and remember.