Askwiseo 🧠📄 — AI-Powered PDF to Knowledge Base System

Askwiseo transforms static PDFs into smart, queryable knowledge bases using powerful AI models. Built with a modern stack and optimized for speed and usability, Askwiseo is your go-to solution for turning documents into interactive, searchable experiences.

---

🚀 Live Demo

Coming soon!

---

✨ Features

• 📄 Upload and parse PDFs into structured content
• 🔍 Query the knowledge base in natural language
• 🤖 Powered by advanced LLMs (OpenAI / Hugging Face)
• 🧠 Semantic search with vector embeddings (Pinecone)
• ⚙️ Fast and scalable backend (FastAPI)
• 💾 Supabase for user management and document storage
• 🌐 Sleek and modern UI built with Next.js & Shadcn/UI

---

🛠 Tech Stack

Frontend:

• Next.js (App Router)
• Tailwind CSS + shadcn/ui
• TypeScript

Backend:

• FastAPI (Python)
• LangChain (for LLM integration and agent logic)
• Pinecone (vector database for embeddings)
• Supabase (auth + DB for metadata)

AI Models:

• OpenAI GPT-4 or Hugging Face Transformers (configurable)

---

🧬 How It Works (Internal Architecture)

User Uploads PDF 
    | 
    v 
Text Extractor (PyMuPDF / PDFplumber) 
    | 
    v 
Chunking & Embedding Generator (LangChain + OpenAI/HF) 
    | 
    v 
Embeddings stored in Pinecone (vector DB) 
    | 
    v 
User asks a question 
    -> LangChain QA chain retrieves chunks 
    -> LLM generates answer 
    -> Response shown in frontend

Bonus: Metadata is stored in Supabase (file name, upload date, user ID) for traceability.

---

⚙️ Getting Started

1. Clone the repo

git clone https://github.com/your-username/askwiseo.git
cd askwiseo

2. Set up environment variables

Create a .env file for both frontend and backend:

.env (FastAPI)

OPENAI_API_KEY=your_key
PINECONE_API_KEY=your_key
SUPABASE_URL=your_url
SUPABASE_SERVICE_ROLE_KEY=your_key

.env.local (Next.js)

NEXT_PUBLIC_SUPABASE_URL=your_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_key

3. Run the app locally

The frontend calls same-origin /api/* routes. In development, Next.js rewrites those requests to the FastAPI backend at http://127.0.0.1:8000, so both services must run.

npm install
npm run dev

This starts:

• Next.js on http://localhost:3000
• FastAPI on http://127.0.0.1:8000

Make sure backend/firebase-service-account.json exists, or set FIREBASE_SERVICE_ACCOUNT_JSON for the backend, so plan and document data can be read from Firebase.

4. Run the backend only (FastAPI)

cd backend
venv\Scripts\python.exe -m uvicorn main:app --host 127.0.0.1 --port 8000

5. Run the frontend only (Next.js)

npm run dev:web

---

🚀 Deployment

Vercel (Frontend)

• Ensure vercel.json is at the repository root.
• Add SENTRY_DSN secret in Vercel dashboard (Settings → Environment Variables).
• Connect the repo to Vercel; each push to main triggers a new deployment.

Render (Backend + Frontend)

• Ensure render.yaml is at the repository root.
• Add the SENTRY_DSN secret in Render (Environment → Secrets).
• Create a new service from the repo; Render will automatically build the Docker backend and static frontend.
• Both services will be redeployed on every push to main.

Note: The backend validates required production environment variables at startup. Set ENVIRONMENT=production in your deployment environment to enable the checks.

🗂 Folder Structure

askwiseo/
├── backend/
│   ├── main.py
│   └── services/
├── frontend/
│   ├── app/
│   └── components/
├── README.md
├── .env

---

🤝 Contributing

Got ideas to improve Askwiseo? Open a pull request or start a discussion!

---

📄 License

MIT License.

---

🙋‍♂️ Author

Built with 💡 by Arun Kumar
Email: arunkumar.ai.engineer@gmail.com Project: Askwiseo