Rooted

## Inspiration Many bright high schoolers coast through classes without ever learning how to study, only to hit an "academic shock" in college when exams demand deep synthesis rather than passive recall. Re-reading textbook chapters or highlighting notes creates a false illusion of competence—a phenomenon known as the fluency illusion, where passive recognition is mistaken for actual mastery.

We built Rooted to solve this transition crisis. Inspired by Richard Feynman’s famous learning philosophy—that true understanding means explaining a concept simply in your own words—we set out to transform passive study habits into an active, visual learning experience.

## What it does Rooted is a botanical-themed learning journal that evaluates your spoken explanations in real time and maps your knowledge onto a living, Graphviz-generated tree: 1) Explanation: Speak through a tough concept naturally. Rooted transcribes your audio via ElevenLabs or accepts manual typed notes. 2) Cognitive Gap Analysis: If student has understood a concept thoroughly, a branch grows signifying their progress in growing their tree of knowedge. 3) Targeted Resources: To patch identified gaps, Rooted automatically generates custom one-click reference bookmarks for Khan Academy, Wikipedia, YouTube, and Google guides.

## How we built it Frontend & Visuals: Streamlit, Custom CSS (Botanical Field Journal Engine), HTML, Graphviz (DOT Engine)

Backend & Intelligence: Python, Gemini 3.8-flash API, ElevenLabs API, Requests, Web Searching / Resource Linking Engine

## Challenges we ran into A lot of the software used in this project was completely new to our team members, including ElevenLabs, Google Gemini API, and Graphviz. This hackathon challenged us to step outside our comfort zone and tackle new tools, and there was definitely a steep learning curve in the beginning. With the help of workshops, mentors, and AI research, we were able to integrate these technologies into our project.

Going into the technicalities, we had trouble handling API updates and model deprecations on the backend, which were crucial for our evaluation feature to work reliably. Additionally, in the frontend development, we had problems with the alignment of visual components and styling Streamlit to fit our field journal theme, alongside general user interaction with the voice-to-text pipeline. As for the teamwork aspect of this hackathon, we initially had trouble delegating tasks between members but as we made a concrete outline of the components necessary for the project we were able to split the work up evenly and everyone contributed somethign valuable to the project.

## Accomplishments that we're proud of We are proud that we took on a project that pushed us all to grow and step outside our comfort zone, and it was satisfying to see the backend analysis and frontend visual tree come together. We’re also proud that our work aligned with a mission of helping incoming students overcome academic shock and build better study habits for college, being first years who intially experienced this ourselves.

## What we learned We learned about implementing new technologies, more specifically Google Gemini API (gap analysis, intent classification), ElevenLabs API (speech-to-text), and Graphviz (dynamic node trees). None of us had worked with the Gemini API before, so we had to learn how to prompt it within our code to output clean concept lists. With Gemini, we turned raw voice transcriptions into structured nodes that integrated with our frontend to acheive visually intuitive mastery trees. Streamlit and CSS enabled us to manage the user flow and design an instinctive visual interface.

## What's next for Rooted We will introduce personalized pathways to help students with semester roadmaps, topic check-ins, and academic planning. The platform can grow beyond individual subjects by tailoring resources and guidance to specific course syllabi across campuses. To keep students engaged, we’ll add simple gamification features like growth milestones and streak tracking that encourage regular study habits. We will also expand the web app to include a persistent knowledge forest, interactive branch drills for missing concepts, and exportable PDF study journals for exam prep.

Built With

• css3
• data-visualization
• elevenlabs
• feynman-technique
• gemini-api
• generative
• graphviz
• html5
• natural-language-processing
• prompt-engineering
• python
• streamlit
• voice-recognition