Hacker News new | ask | show | jobs
by pghqdev 148 days ago
Built this because I wanted a notes app that:

1. Works offline (airplane, subway, spotty wifi)

2. Doesn't require an account

3. Keeps data on my device, not someone else's server

4. Something to take notes right there in the browser

Technical stack: Next.js static export, IndexedDB for storage, Web Speech API for voice dictation. Total infrastructure cost is ~$0/month for static hosting.

The architecture is intentionally simple. Notes live in the browser's IndexedDB. Voice dictation uses the browser's native speech recognition. No external APIs, no database, no waitlists, no sign-ups

Trade-offs I accepted:

- No sync between devices (considering E2EE sync as optional paid feature)

- Data loss if you clear browser storage (export to MD/PDF available)

- PWA only, no native apps

Who actually uses this:

- Quick capture when you don't want to open a full app

- Drafting prompts before pasting into ChatGPT/Claude (has token counter)

- Private journaling without cloud anxiety

- Offline notes on flights/commutes

- People with "account fatigue" who just want to write

Curious what HN thinks about local-first as a default architecture choice. The privacy angle resonates with users, but the no-sync limitation is a dealbreaker for many.

Happy to discuss technical decisions or answer questions.

1 comments

How do you get this if there's "no external APIs"?

> Text-to-Speech > Listen to your notes with built-in text-to-speech. Choose from multiple TTS providers and voices to suit your preference.

> 3 TTS providers: Browser, ElevenLabs, OpenAI

That is an optional + feature in the upcoming roadmap where users can choose to BYOK for TTS if they want to.

In the core product we are not using the like of either ElevenLabs or OpenAI at all