Back to projects

Project

Personal Knowledge Base

A central platform for developers and knowledge workers to capture, search, and retrieve work context across notes, audio, and OCR.

MVP LLM ClassificationOCRVoice Transcription

Problem

Developers and knowledge workers accumulate context faster than they can organize it. Important details live across calls, chats, screenshots, notes, and task tools.

Why I Built It

Frequent context switching reduced productivity and caused mental fatigue. The goal was to create a trusted system that could capture context with minimal friction and surface it when it mattered most.

Architecture

The initial architecture separates capture, normalization, indexing, retrieval, and answer generation. Notes, audio transcripts, and OCR text become normalized records before embedding and search.

Tech Stack

  • React Native
  • Postgres
  • Embeddings
  • RAG

Current Status

Growth and marketing. The first milestone was to build a narrow capture-and-retrieval loop before expanding into richer ingestion.

Next Steps

Feature development will be guided by user feedback, with each iteration shaped by real usage patterns and pain points.

Further Reading

Read related technical notes.

The writing archive expands on the architecture patterns, product constraints, and AI workflow decisions behind these projects.