Frank.
Frank.
PROJ.03Privacy-First AI Workspace

Noteable

A local-first, privacy-focused AI notepad that runs entirely on-device, offering instant voice capture, smart summarization, and interactive chat without cloud exposure.

System StatusLive
Deploy Year2026

Architecture Map

NODE.01Next.js
NODE.02React.js
NODE.03Transformers.js
NODE.04IndexedDB
NODE.05Tailwind CSS
Phase 01

The Challenge

Strategic Problem
Statement

Modern AI applications typically rely on remote cloud APIs, posing serious data privacy risks and latency issues. The goal was to build a highly responsive workspace where document processing, semantic auto-linking, and voice synthesis happen entirely offline.

Phase 02

The Solution

Engineering
Execution

Developed a local-first architecture leveraging Transformers.js and Web Assembly for client-side LLM inference. Integrated IndexedDB for secure offline storage, and built a custom Web Audio pipeline for real-time transcription and fluid voice capturing.

My Role:Lead Full-Stack Engineer

Outcome Report

A local-first, privacy-focused AI notepad that runs entirely on-device, offering instant voice capture, smart summarization, and interactive chat without cloud exposure.

  • Achieved 100% on-device data privacy with zero cloud exposure or external API requests
  • Engineered 'Note Buddy', an on-device AI chat assistant that dynamically queries and connects local notes
  • Implemented a real-time speech transcription pipeline ('Fluid Capture') converting audio inputs directly into structured notes
Core Metric
100%
On-Device
Initiate Similar Build