Drawer stores readable files on your disk. Raw sessions stay local, and changing tools never takes your memory away.
- Local-first
- Bring your own key
- Portable files
AI keeps getting smarter, yet it still doesn't know you. Every new session, every new tool — you introduce yourself all over again, while your preferences, projects and hard-won lessons scatter across a thousand chats.
You've said "save it to Desktop, date goes first" a hundred times. You'll say it again tomorrow.
Last month's lesson gets relearned in next month's tool. Experience stays anecdote — it never becomes an asset.
When an app does remember you, the memory lives in their cloud — unreadable, unportable, revocable.
Drawer keeps durable facts and repeatable actions together. Watch one project become more useful without losing where each change came from.
01 / 03
Files, notes and preferences arrive from different places. Drawer links them to one object without flattening their sources.
Project object · owned by you
Give Drawer ordinary language. It proposes structure, separates lasting facts from repeatable work, and waits for your approval before anything is saved.
Edit the text or choose an example, then generate a reviewable object. You can approve it or change the note and try again.
Interactive local preview · edit the note, generate a proposal, then approve it to see the saved state.
Drawer appears over the tool already in front of you. Find a behavior, press Enter, and control passes straight back to your agent.
Drawer isn't another filing chore. Your drawers grow out of the work you already do with AI.
One identity through every revision
A reviewed project brief becomes one portable object with a clear owner and scope.
Drawer stores readable files on your disk. Raw sessions stay local, and changing tools never takes your memory away.
Attributes and behaviors remain drafts until you approve them. Drawer records the decision instead of silently changing what an object knows.
The panel appears only when called, passes the selected behavior to your agent, then gets out of the way. No permanent workspace competes for attention.
A standard plain-text format keeps objects independent from any one model or agent host. Change the tool; keep the same reviewed memory.
Bymii AI Global Limited is developing Drawer, a user-controlled memory and workflow layer for AI tools. It helps people organize projects, preferences and recurring work into portable objects they can review, retain and use across compatible AI environments.
How AI assists: Drawer uses AI to extract structure from user-provided information, draft attributes and suggest reusable behaviors. Suggestions remain subject to user review and approval.
Our company, our projects, our lives — organized as objects, run with AI. Early access opens soon. Be the first to know.
Get early access