The first release covers signup only.
A private context layer for your Mac.
Pick up where
you left off.
We’re building modalis to read the files and AI chats you choose and keep one page per project on your Mac, each line linked to its source, so your next AI chat can start from that page instead of from zero.
For Mac · in development

Your project is more
than your last chat.
The brief is in a folder.
The decision is in a Claude chat.
The next step is in Codex.
Every new chat starts from zero.
Ship Friday, as long as QA is green.
Fix the auth redirect before launch.
Who signs off the final copy?
Meet the Librarian.
Here’s the Librarian we’re building. It reads only what you
point it at and keeps each project written down, with receipts.
“Only what you hand me.”
You pick. It reads nothing else.
Tick the folders, repos and Codex or Claude Code chats that matter. Everything else stays out.
4 of 5 feed the Harbor page · tap one to switch it
A longer memory.
Still your say.
Lives on your Mac.
The Librarian reads and writes on your machine, so its own work has no per-token bill.
Shows its sources.
Each line links back to the file or chat it came from. Your fixes stay marked as yours.
Asks first.
You choose what it reads, and you decide whether an assistant can use the page.
Good things
take context.
We’re building toward one useful moment:
reopening a project and knowing where to begin.

Before you join.
Can I use it today?
Not yet. It’s in development and there’s no release date. Join the waitlist and we’ll email you when you can try it.
Which AI chats can it read?
Codex and Claude Code sessions on your Mac, and only the ones you choose. Nothing else for now.
Does my work stay on my Mac?
By default, yes: your files and its pages stay on your Mac. If you connect an assistant, that assistant’s own rules apply to what it reads.
Is it another AI chat app?
No. It keeps a plain, sourced page for each project first. Then, if you allow it, your assistant works from that page.
