Notionary

Context

Students don't trust AI for study, because it confidently gets course material wrong. Notes get passed round on WhatsApp and are gone by May, and the slides never capture everything a lecturer says.

Notionary was our answer at the Notion AI Hackathon during Dublin AI Week: your class builds one shared course knowledge graph, and your AI uses it to teach you personally.

What we built

• A shared course space in Notion, with sources and linked concepts.
• Two students asking the same question and getting different explanations, such as a worked example for one and a harder practice question for the other, both citing the same source.
• A revised assignment announcement that updates the deadline for everyone, with the revision as its source.
• Private learning profiles that stay separate from the shared space.

My role

I built the backend: the Supabase functions, the Notion integration, and the Claude ingest that turns new sources into linked concepts. Spencer built the frontend, and King and Marly wrote the pitch and built the slides.

Built with Spencer Kesse, King Mbachu and Marly.

How it went

We had a little over three hours to build it, and it won Best Use of Notion.