Privacy Promise
journalingtechniques.ai was built for people who refuse to compromise on privacy. Our entire architecture makes it mathematically impossible for us to touch your journal text.
How it works
Your Browser β ββ Loads static Next.js site from CDN β ββ Downloads Hugging Face models (once) via Transformers.js β ββ Runs emotion + theme analysis locally (WebAssembly/WebGPU) β ββ Shows insights + CBT question Zero API calls with your text. Only optional telemetry: helpful/not helpful clicks.
What we log (and what we never will)
We collect
- β’ Prompt type (daily reflection / CBT / gratitude)
- β’ Word count + character count
- β’ Whether insights were helpful (π/π)
- β’ Model download duration (for performance tuning)
We never collect
- β’ Raw journal text or generated insights
- β’ IP addresses, cookies, device IDs
- β’ Names, emails, or any identity markers
- β’ Anything that could be reconstructed into your story
Trust checklist
- AI models download to your browser via HTTPS and cache in IndexedDB
- Journal text never leaves memory β there is no server or API receiving it
- Analytics collect anonymous metadata only (word counts, prompt types)
Need extra assurance? Open DevTools β Network tab and watch the silence while you journal. Or email privacy@journalingtechniques.ai for a full threat-model walkthrough.