I have a problem that I think a lot of people here share: I use too many AI platforms.
ChatGPT, Claude, Gemini, Deepseek, Copilot, Perplexity, Mistral, and a handful of others depending on which one is having a good week, which one just raised rate limits, and which one hasn't gone down in the last 24 hours.
The actual AI part is fine. The problem is afterwards. I'd have a conversation three weeks ago where I worked through something useful, a coding approach, a content structure, a strategy I wanted to revisit, and I'd have no idea which platform it was on. I'd open tabs, scroll through histories, try half-remembered keywords in four different search bars, and usually just give up and start the conversation from scratch.
I kept thinking someone must have built a tool for this. Nobody had. So I built a simple thing to help.
LLMnesia is a Chrome extension that indexes your AI conversations across platforms and lets you search them all from one place. Everything stays local on your machine. Nothing is sent anywhere. No account required.
I've been using it daily for over a month now and it's become the starting point for most of my AI work. Finding any previous thread takes under a second instead of the five-minute tab-opening ritual I was doing before.
Getting it into the Chrome Web Store was its own mini adventure. First submission got rejected for keyword spam in the description, which I didn't realise was a thing. Rewrote the whole listing, resubmitted, approved a few days later.
It's free. No premium tier, no plans to gate features.
If you work across multiple AI platforms, I'd genuinely appreciate you trying it. Specifically: does the indexing pick up your platforms? Is the search finding what you expect? Anything feel off or missing? I'm building this solo and I'll ship fixes based on whatever comes back.
https://chromewebstore.google.com/detail/leekfgbdojiaabifbjbbgiiclannjdkf
The “which tab was that in” tax is worse than the model being wrong. I stopped treating any one sidebar as the archive — when a thread actually lands I copy the prompt + a one-liner on why, so the next session isn’t a scavenger hunt across ChatGPT/Claude/Gemini. Do you still start from scratch when you can’t find it, or keep a tiny win-log outside the chats?
This is too real — the “which platform did I use?” problem is honestly worse than the AI itself 😅
Love the local-first approach, that’s a big trust win. And the use case is super clear.
Curious how well it handles different platform formats over time — feels like that’ll be the tricky part.
Also, if you’re building tools like this — $19 puts it in real competition. Tokyo trip + $500 min guaranteed.
Round just opened: tokyolore.com 🚀