Hi everyone, I’m the developer behind Study Assistant. I built this because I needed a way to query my collection of study documents without relying on cloud-based search or proprietary interfaces. It’s a local-first RAG desktop application designed for privacy and speed. Key Technical Highlights: • Local-First Architecture: Runs fully offline via Ollama. • Fallback Chain: Implements a multi-provider strategy (Ollama → Gemini → Groq → OpenRouter) to ensure query success even if local compute capacity is hit. • Smart Indexing: Uses file-hash validation to cache embeddings, preventing redundant processing of your document library. I’m specifically looking for technical feedback on the retrieval accuracy and the UX of the multi-provider fallback logic. The code is entirely open-source and available for audit. Looking forward to your technical critiques and suggestions for improving retrieval latency.

Study Assistant

Privacy-first app to chat with your local documents.

Privacy-first app to chat with your local documents.