Personal AI Memory
Captures and stores your chat from various AI platforms
Start new thread
Mars Wang

Personal AI Memory V2 - Captures and stores your AI Conversation from LLM platforms

by
Personal AI Memory
•
AI Memory is a privacy-first browser extension that captures, organizes, and lets you seamlessly recall your ChatGPT, Gemini, Perplexity, Grok, and Claude conversations locally without data leaks.

Add a comment

Replies

Best
Mars Wang
Maker
📌
Hey Product Hunt! 👋 I'm Mars, the solo developer behind Context Vaults. Today, I’m thrilled to launch the V2 update, bringing Grok, Perplexity, and Claude into the ecosystem alongside ChatGPT and Gemini. 1. What inspired you to build this? As an engineering grad student constantly researching complex system architectures and writing code, my daily workflow is scattered. I use ChatGPT for initial brainstorming, Perplexity for deep-dive research, and Claude for actually writing and refactoring the code. The inspiration hit me when I realized how much time I was wasting just trying to keep all these different "brains" on the same page. I wanted one unified memory layer that sits across all of them. 2. What problem were you trying to solve? The "Copy-Paste Hell" and Context Isolation. If I spend 30 minutes explaining a specific software architecture to Perplexity, and then switch to Claude to write the implementation, Claude knows absolutely nothing about it. I was burning tokens and losing my train of thought constantly copy-pasting context windows. Furthermore, existing solutions required sending my private chat logs to a 3rd-party cloud, which was a massive privacy dealbreaker for me. Initially, I just wanted a simple script to pass text between tabs. But as I dug deeper into the technical design, the privacy aspect became my absolute priority. My approach evolved from a basic bridge into building a robust, 100% local-first system. For this specific V2 launch, integrating Grok, Perplexity, and Claude was a huge technical hurdle. I’d love for you guys to tear it apart, test the local hybrid search, and give me some brutal feedback! Let me know what you think. 👇
Kunal Raha

@marswangyang Hey Mars — love the local-first approach, privacy was clearly the right call.

Feedback: The cross-LLM context retrieval is genuinely the hardest part of this problem. If I start a conversation in Perplexity and continue in Claude, the embedding models are different — Perplexity might use one model, Claude another. Your hybrid search needs to normalize across those embedding spaces or relevance scores will be inconsistent.

Question: How are you currently handling embedding model inconsistency across platforms? Are you re-embedding everything through one unified model locally?"