Discover Memoriq, an open-source, E2E encrypted private vault to save and protect your AI conversations. Self-hostable and ultra-secure.
Ever spent 20 minutes scrolling through your endless ChatGPT history looking for that one killer prompt you wrote last month, only to realize OpenAI probably fed it to their next model anyway? Or maybe you just got that cold sweat running down your spine thinking about how much sensitive company code you’ve copy-pasted into Claude?
Well, you are not alone in this paranoia. A new tool called Memoriq just landed on Product Hunt, bagging a solid 84 score. It promises to be an end-to-end encrypted (E2EE), private AI memory vault for all your chats across ChatGPT, Claude, Gemini, and Grok. Even better? It’s completely open-source and self-hostable.
In plain English, Memoriq is a personal knowledge vault. Instead of leaving your valuable AI chats scattered across different platforms—or trusting another "we promise we are secure" plaintext cloud service—Memoriq lets you hoard them all under your own lock and key.
According to the creator, the project was born out of pure frustration from losing useful conversations when hopping between LLMs, combined with a healthy distrust of standard plaintext cloud storage.
The key highlights:
As soon as Memoriq hit the front page, the crypto-nerds and system architects on Product Hunt wasted no time putting it under the microscope.
One sharp dev immediately asked about the cryptographic primitives being used and how search doesn't crawl to a halt.
The creator didn't flinch, laying out the practical architecture: "The current approach is to keep the cryptography conventional and avoid clever encrypted indexing schemes for now. Memoriq uses a browser-generated 256-bit AES-GCM master encryption key (MEK). The user's password is processed with PBKDF2-SHA256, and the MEK is wrapped with AES-KW."
To avoid the search bottleneck, the server does absolutely zero content searching. When you unlock your vault, only the encrypted conversation headers are decrypted locally in your browser to build the list and search index, while full conversation bodies are decrypted on-demand. Practical, fast, and secure.
Another great question touched on the capture mechanism. Since LLM interfaces are vastly different, how reliable is the data gathering?
The creator admitted it's a bit of a cat-and-mouse game:
Can we use these chats to feed other LLMs or export them easily?
Right now, exports are just encrypted backups. However, the creator’s long-term vision includes semantic search, privacy-preserving AI agents that can access your private conversation library, and end-to-end encrypted sharing.
Let’s be real here. As senior devs who have been burned by "secure cloud" promises more times than we can count, Memoriq’s approach is refreshing.
Sure, the DOM scraping mechanism for Claude and Gemini is bound to break whenever their front-end devs feel like pushing a random UI update on a Friday afternoon. But hey, it's open-source—when it breaks, we can just submit a PR.
If you are a privacy enthusiast who loves hoarding AI prompts and doesn't mind spending 15 minutes of your weekend configuring your own setup, Memoriq is an awesome weekend project. But if you're too lazy to care and don't mind OpenAI using your bad code to train GPT-5, just stick to the default web UI and call it a day.
Source: Product Hunt
