filed under: open loops
One memory. Every AI.
Your private memory system connecting Claude, ChatGPT, Codex, or Antigravity to Cursor, Lovable, Replit, or Bolt.14 days free · no credit card required
no folders. no formatting.
You work. It remembers. The decisions, fixes, and trade-offs you make while working are captured automatically, with the why attached.
filed in claude. recalled in cursor.
You will forget the fix.
the forgetting curve · unassisted · day 112
It won't.
“add access-control-allow-origin to the worker response, that took a whole evening”
held · permanently
Working Memory hands your AI the exact context it needs, not a dump of old chat logs. You stop spending the first five minutes of every chat re-explaining your work.
reason in one ai. build in the next.
what we refuse to build
What it won't do.
You didn't file it. You didn't tag it. It's there anyway.
No training on your notes
Your captures are yours. Encrypted where they're stored, never used to train any model. Export your data any time.
No always-on mic
“Ambient” means it works from what you're already doing in your AI, not a microphone. Nothing in the room is heard or watched.
No buckets at the point of capture
You'll never be asked “which list does this go in?” while you're trying to remember the thought.
No badges, streaks, or guilt
No pressure. Just the simple act of capturing your thoughts.
One surface, not seven views
Capture, query, and recall share a single thread. No Kanban. No calendar grid. No left sidebar.
Not another chatbot
You're not writing prompts for a bot. You're leaving notes for a future version of yourself.
the old way costs more the more it knows
It won't cost more because it knows more.
up to90%
fewer tokens of context per conversation than keeping your notes in the prompt
Measured across Claude Opus 4.8, Sonnet 5, and Haiku 4.5. The more your AI remembers, the more you save. Fewer tokens means a lower bill, and longer before you hit a limit.
Token costs increase with the size of your data, and your LLM can't remember everything from every chat. Your Working Memory stays at one low price no matter what.
one line item, forever
One plan. Everything in it.
Unlimited capture, recall, and storage across every AI you connect.
$19.99/ month
- Unlimited capture, recall & storage
- Connect every AI you work in
- Recall across your entire history
- Classification & search built in
no tiers · no seats · no meetings
Start free trial14 days free · no credit card required
asked and answered
Do I have to save anything?
No, that's the whole idea. As you work in your AI, Working Memory notices what's worth keeping and hands it back when it's useful. No saving, no filing, no deciding which list a thought goes in.
Is this just another AI notes app?
A notes app gives you a blank page and asks you to do the work of organizing. Working Memory does the organizing for you, in the background. The AI is behind the glass, not in front of it.
What happens to my notes?
Working Memory is privacy-first. Your notes are encrypted for your eyes only, never used to train any model, and exportable at any time. If you cancel, you can continue to access your data for one year.
How is this different from Notion / Obsidian / others?
Tools like Notion or Obsidian ask you to stop working to organize folders and format documents. Working Memory lives inside your AI tools, remembering context in the background so you stay in your flow.
Where are the native apps?
Working on it! In the meantime, Working Memory lives inside the AI you already work in. Connect it to Claude, ChatGPT, Codex, Cursor, Google Anti-Gravity, and more and it keeps up from there.
Can I cancel?
Any time, one tap, no questions asked. If you cancel, your notes stay with you. We hope you won't want to, but if you do, we won't make it hard.
Why are you charging for Working Memory?
A privacy-first experience like this isn't free to build or maintain. We offer a 14-day free trial so you can see if it's right for you, and we hope the value it provides will speak for itself. If you have feedback on pricing, say hi!
the humans behind the glass
Built by parents who are tired of repeating themselves.
Working Memory exists because you shouldn't have to re-introduce yourself to every AI, every morning. Re-pasting the same context into five different chats is a ridiculous way to spend the 21st century. Working Memory is the cobrain they all share: it fills itself in from your work, and every AI you work with just knows.
If you have ideas, we'd love to hear them.