Proxima is a modern, responsive, and friction-less Retrieval-Augmented Generation (RAG) chat application built with Blazor Interactive Server. It provides visitors with a premium, drop-in experience to upload documents and instantly converse with an intelligent AI aware of their custom context.
- Anonymous Sessions: Instantly tracks user chats and documents without requiring logins or signups utilizing secure HTTP-only cookies.
- Drag-and-Drop Document Ingestion: Upload
.txt,.pdf,.md, and.docxfiles through an intuitive Drag-and-Drop responsive UI. - Multi-Conversation Persistence: A built-in SQLite repository securely saves, categorizes, and restores unlimited distinct chat histories per session isolation block.
- AI Auto-Titling: Automatically generates concise summary titles for your active communication threads using lightweight, background LLM queries.
- Responsive Mobile Flyout Sidebar: Engineered to guarantee a pristine Desktop side-push UI that effortlessly converts to an absolute, touch-friendly Flyout on mobile displays.
- Advanced RAG Engine: System-prompted to prevent hallucination. Responses strictly rely on uploaded context utilizing intelligent Semantic Search vectors.
- Framework: C# 13 / .NET 9 ASP.NET Core Blazor (Interactive Server)
- Database: SQLite (
Microsoft.Data.Sqlite) - Semantic Vector Store:
Microsoft.Extensions.VectorData - Document Processors:
DocumentFormat.OpenXml(Word), PDF/MD Readers - Styling: Pure CSS (Dark Mode Glassmorphism & Desktop-Mobile Media Queries)
- Clone the repository and navigate to the project directory.
- Restore any necessary NuGet packages:
dotnet restore
- Build and execute the application:
dotnet run
- Navigate to the local URL (usually
http://localhost:5215) to immediately start chatting with your uploaded documents.