I built Memory Sync so I don't have to reteach every AI who I am

I built this because I kept using different AI tools for different things, and got pretty tired of re-explaining the same stuff over and over.

ChatGPT knew a few things about me, Claude knew some other things, Gemini had basically none of it, and switching between them always felt a bit dumb.

So this is my attempt at a portable memory layer.

Right now the idea is pretty simple:

• keep one Memory.md as the source of truth

• pull memory out of one platform

• clean it up / edit it

• push it into another one

• show the last pull / push time for each platform, so you can actually tell whats up to date

It currently works with ChatGPT, Claude, Gemini, Grok, Kimi, Mistral, and Copilot.

One thing thats intentional: its not fully automatic. The extension opens the target site, injects the sync prompt, and tracks sync state / last sync times, but you still review and send things yourself. I wanted the first version to be more transparent and controllable, not "magic" but hard to understand.

I also kind of think markdown is a decent starting point here because its:

• editable

• portable

• easy to back up

• not tied to one company

Still pretty early, and I’m not totally sure what the right long term format is. Could end up being:

• plain markdown

• markdown + some structured fields

• or a proper import/export format if this space matures

Curious if other people here have the same problem, or if this is mostly just an AI power-user thing. also very interested in pushback.