Project Structura

Document Intelligence Reimagined Using AI

Comment

Inspiration

We were inspired by the chaos of unstructured data. We wanted to bridge the gap between messy documents and usable assets, creating a tool that can instantly turn a rough draft into a polished Kindle eBook, a complex concept into a Mind Map, or a PDF into a queryable database.

What it does

Structura AI transforms documents (PDFs, DOCX, Images, Code) into structured data and visual formats. It offers specific tools for: Refinement: Grammar fixing and professional rewriting. Visualization: Generating automatic Flowcharts, Mind Maps, and ER Diagrams. Data Extraction: Converting text to strict JSON or CSV. Ask AI: A flagship mode to chat with documents, generating deep insights and suggested questions.

How we built it

We built the application using React and Tailwind CSS for a modern, glassmorphic UI. The intelligence is powered by Google's Gemini API (specifically Gemini 3 Flash and Pro models). We utilized pdfjs-dist and mammoth for client-side file parsing and mermaid.js to render AI-generated diagrams dynamically.

Challenges we ran into

The biggest challenge was Prompt Engineering to ensure the AI returned strictly valid code (JSON/Mermaid) without conversational filler. Handling client-side PDF parsing for large files and ensuring the "Ask AI" chat context remained accurate within token limits were also significant hurdles.

Accomplishments that we're proud of

We are proud of the Visualizations engine, which seamlessly turns text descriptions into complex, interactive diagrams. We also love the Ask AI interface, which combines a document preview (including PDF rendering) with an interactive, context-aware chatbot in a single cohesive view.

What we learned

We learned how to leverage System Instructions to enforce specific personas (e.g., "Corporate Editor" vs. "Data Engineer"). We also gained experience in handling binary file data in the browser and managing asynchronous AI streams to keep the UI responsive.

What's next for Project Structura

Full Offline Mode: Integrating WebLLM to run small LLMs entirely in the browser (WebGPU) for total privacy. Batch Processing: Allowing users to upload 50+ files at once for bulk extraction. API Gateway: Offering a developer API so other apps can use Structura’s cleaning engine.

Built With

Share this project:

Updates