Changelog

Launched Jadebook OSS version. It’s probably one of the most privacy-friendly journal platforms on the market since you setup your own database, and host everything on your own accounts.

Check out the GitHub →

Benefits

• Everything is held in Supabase, including auth, db and storage

• Get search, tags, covers, icons, transcription and uploading assets

• Mobile-responsive and available as a Progressive Web App (PWA)

• Comes with OpenAPI spec in-case you want to build something on top

• Uses NextJS so the code is fairly accessible and easy to work with

• Export/Import to and from the SaaS Jadebook platform

• You can one-click deploy it on Vercel for free

• Rich-text editor for entries

When we first built Jadebook, we built it in a monolith-structure where everything was in one place. While this worked great, the scope of Jadebook has increased — where we want to support different platforms. The previous architecture made this extremely difficult so we completely reworked it.

Here's what's been shipped:

1. Flat 20-30% performance improvement

2. Increased backend security and reliability

3. Memory is now much faster to generate and more reliable

4. Major UI overhaul to make everything less janky and far more versatile

5. The editor is now simpler due to our plans to build a mobile app

Oh and we made the marketing site way better :)

We plan on launching on Monday, 2nd June. As such, we've shipped a bunch of updates including many platform bug fixes.

• Use Gemini 2.5 Flash over Gemini 2.0 Flash

  • We've disabled thinking output but in general, 2.5 Flash is faster, more reliable, has more knowledge and has more capabilities

• Allow users to include images and PDFs to their chats

• Updated some document layouts to work more intuitively

• Separate journals

  • We now have the home view which shows all entries from all types of journals.

  • Journals now have a separate section in the sidebar

• Update all empty states to be more friendly to new users

• Add voice transcription (BETA)

  • Free users can use real-time browser transcription if supported

  • Pro users can record in 10-minute sessions and have the OpenAI Whisper model generate the transcription. This will also add grammar and is way more accurate

• Update Memory structure to be more reliable for models to generate — also added goals and aspirations and Values to the schema

• Updated the model used for Memory

  • Previously, generation took take up to 15 seconds. It should now take ~1 second with ~5 seconds being the slow end

• Allow new users to quickly go through an onboarding process, setting their desired theme, username, and layout

• Create onboarding document — shows new user's the rich-text capabilities and AI features

• Add the final background functions for handling automatic document deletions (for entries that have been deleted for over 30 days)

• A metric ton of bug fixes