Cloud AI tools are powerful — but sometimes, you don’t want to upload your private files.

That’s why I created a Blazor Server app that lets you chat with your PDFs offline using Ollama. It’s fast, private, and free.

📂 GitHub Repo: https://github.com/DheerGupta35959/BlazorPdfChat

Key Features

• 📄 Upload multiple PDFs
• 🧠 Automatic text extraction & chunking
• 💾 Local vector storage with LiteDB
• 🔍 AI answers only from your documents
• 💬 Realtime streaming chat with token-by-token updates

Architecture

Blazor Server (UI) → ASP.NET Core API → LiteDB Vector Store
                         ↕
                   Ollama REST API
                         ↕
                    Local AI Models

Tech Stack

• .NET 8 / Blazor Server
• PdfPig → Extract PDF text
• LiteDB → Store embeddings & chunks
• Ollama → Embeddings (nomic-embed-text) + Chat (llama3)
• SignalR → Stream AI responses

How It Works

1. Ingest PDFs

• PDFs are parsed into plain text
• Text is split into manageable chunks
• Each chunk is converted into embeddings via Ollama
• Chunks + embeddings are stored locally in LiteDB

2. Chat

• User enters a question in the Blazor chat
• The app searches LiteDB for the most relevant chunks using cosine similarity
• Context + question are sent to Ollama’s chat model
• Tokens are streamed back in real-time

Setup Guide

1. Install Ollama → https://ollama.ai

2. Run Ollama in the background:

   ollama serve
   

3. Pull models:

   ollama pull llama3
   ollama pull nomic-embed-text
   

4. Clone the repo:

   git clone https://github.com/DheerGupta35959/BlazorPdfChat.git
   cd BlazorPdfChat
   

5. Run the app:

   dotnet run
   

6. Open http://localhost:5000 in your browser
7. Upload PDFs → Start chatting

Why This Matters

• 🚫 No API costs
• 🔒 Works completely offline
• 📈 Perfect for internal docs, contracts, research papers

Conclusion

AI doesn’t have to mean giving up privacy or control.
With Blazor + Ollama, you can have the power of AI on your own terms.

📂 GitHub Repo: https://github.com/DheerGupta35959/BlazorPdfChat