StudySmith

Upload. Learn. Master — in seconds.

Comment

StudySmith Project Story

Inspiration

As students, we often spent hours manually turning lecture slides, PDFs, and notes into actionable study materials — creating flashcards, summarizing content, and trying to extract meaningful learning objectives. It felt inefficient and overwhelming. We wanted a way to automate the entire process so students could focus on learning rather than busywork. The inspiration came from the idea: “What if your notes could teach you back, automatically?”

What it does

StudySmith transforms raw lecture materials into structured learning experiences.

  • Upload your notes, slides, or PDFs.
  • Instantly generate measurable learning objectives using Bloom’s taxonomy.
  • Create smart flashcards with spaced repetition for active recall.
  • Ask questions and receive cited answers directly from your notes.
  • Search by meaning or keyword using semantic + full-text search.

All of this is powered by TiDB, which serves as the backbone for storing embeddings, retrieving content semantically, and connecting all modules in a fast, scalable workflow.

How we built it

We built StudySmith as a Next.js + TailwindCSS frontend with Framer Motion animations for a polished, interactive experience.

  • File Ingestion: Users upload PDFs, DOCX, PPTX, or plain text files.
  • Vector Storage & Retrieval: TiDB Serverless stores embeddings of all uploaded content, enabling fast, semantic searches and Q&A.
  • Objective & Flashcard Generation: A backend pipeline generates learning objectives and flashcards, connecting results to the original notes.
  • Multi-Step Automation: TiDB acts as the central memory, linking uploads → analysis → flashcards || Q&A || Quizes seamlessly.

TiDB made it possible to scale effortlessly, store and query vector embeddings, and retrieve precise answers in real time — all without complex server management.

Challenges we ran into

  • Diverse Document Formats: PDFs, slides, and Word docs all have different structures; parsing them consistently was tricky.
  • Semantic Accuracy: Creating meaningful learning objectives and flashcards required fine-tuning the AI pipeline.
  • System Integration: Coordinating multiple modules into one smooth workflow was challenging.

Accomplishments that we're proud of

  • Fully automated upload-to-study workflow in under a minute.
  • Accurate semantic search and cited Q&A that students can trust.
  • Seamless multi-step automation powered entirely by TiDB, linking storage, retrieval, and AI processing.
  • A polished, interactive UI that encourages consistent use and makes learning fun.

What we learned

  • Multi-step AI workflows work best when there’s a reliable memory layer — TiDB enabled this with serverless scalability and vector storage.
  • Handling diverse note formats requires both preprocessing and smart extraction.
  • User experience is just as important as AI accuracy — students need to feel the system is fast, intuitive, and trustworthy.
  • Cloud-native databases like TiDB make scaling and multi-user management painless.

What's next for StudySmith

  • Integrate more content types: videos, audio lectures, and online articles.
  • Enhanced analytics: give students insights into learning progress and gaps.
  • Collaboration features: allow sharing study objectives and flashcards within study groups.
  • Further AI personalization: dynamically adjust study plans based on student performance.

TiDB will continue to be the backbone of these innovations, enabling real-time, scalable, and reliable workflows for all student data.

Built With

Share this project:

Updates