A TidalHACK '26 Creation 🛠

BureauBuddy is a document‑processing application for government and legal forms. It provides PDF ingestion, structured analysis, translation, text‑to‑speech synthesis, and actionable next‑step guidance through a FastAPI backend and React frontend.

Features

• PDF upload with server‑side extraction and analysis.
• Plain‑language summaries and selection‑based simplification.
• Deterministic “next steps” guidance for completed forms.
• Translation of selected content to a target language.
• Text‑to‑speech synthesis via ElevenLabs.
• Minimal UI focused on reading, selection, and playback.

Tech Stack

• Backend: FastAPI, google‑genai (Gemini), ElevenLabs SDK, httpx, pypdf, python‑multipart
• Frontend: React 18, Vite, TailwindCSS

Project Structure

• backend: FastAPI app and API routes
• frontend: React UI

Setup

1) Backend

Create and activate a virtual environment, then install backend dependencies:

python -m venv venv
source venv/bin/activate
cd backend
pip install -r requirements.txt

Create a .env file in the project root:

GEMINI_API_KEY=your_gemini_api_key
ELEVENLABS_API_KEY=your_elevenlabs_api_key
ELEVENLABS_VOICE_ID=optional_voice_id

Start the API:

uvicorn backend.main:app --reload --port 8000

2) Frontend

cd frontend
npm install
npm run dev

The frontend expects the API at http://localhost:8000.

API Endpoints

• POST /analyze_doc/upload: Upload a PDF and extract summary + key information
• POST /simplify: Simplify selected text
• POST /translate: Translate text to a target language
• POST /tts: Generate audio from text (ElevenLabs)
• POST /next_steps: Generate next‑step guidance
• POST /important_info: Extract important details

Python Dependencies

Declared in requirements.txt:

• google‑genai
• elevenlabs
• httpx
• pypdf
• python‑multipart

Notes

• GEMINI_API_KEY is required for all Gemini‑backed endpoints.
• ELEVENLABS_VOICE_ID is optional; a default voice is used if not provided.

License

MIT