July 31, 2026
Memory lane or privacy pain?
Show HN: Shared memory graph for Claude and ChatGPT, over MCP
A new AI memory app drops, and the comments instantly spiral into trust issues and name jokes
TLDR: Universal Memory Layer wants to let different AI chat tools share one private memory about you, so you only have to tell them things once. Commenters immediately split between interest and suspicion, roasting the branding, mocking the sales copy, and asking the obvious question: why should anyone trust an app with their personal history?
A new project on Hacker News is pitching a big promise in very simple terms: tell one chatbot something once, and all your other chatbot tools remember it too. The product, called Universal Memory Layer, wants to act like a private notebook for your AI helpers, storing the important stuff about you in one place so different apps can reuse it later. The creators say it’s visual, editable, private by default, and not just a giant dump of your chat history.
But honestly? The real fireworks were in the comments. One of the loudest reactions was pure "absolutely not" energy: why would anyone hand over their “memories” to yet another AI service? That skepticism landed fast, with privacy jitters and a big side-eye at the whole concept. Another mini-drama was the name itself: calling it UML triggered instant groans from readers who clearly have scars from old-school diagram software and didn’t love the branding choice.
Then came the snark. One commenter roasted the site’s polished, AI-flavored copy, saying it reads like a machine trying to sell a machine. Another dropped the classic startup-killer line: great idea, no one’s ever tried that before, which is basically Hacker News shorthand for “cool pitch, but prove it.” Still, not everyone came to throw tomatoes — at least one reader was genuinely impressed and called it a fresh idea. So the mood was split: part curiosity, part eye-roll, part please don’t upload my brain to the cloud.
Key Points
- •Universal Memory Layer is presented as a private shared memory graph for AI tools, apps, and MCP-connected clients, with free early access and open-source positioning.
- •The product describes a workflow in which users or apps save information, UML extracts durable facts, verifies them in the backend, and stores structured memory instead of raw chat logs.
- •UML can be connected to tools such as Claude, Claude Code, and Cursor through MCP links, and also exposes direct API and SDK integrations.
- •The article states that MCP-based integrations depend on the host client or model deciding when to call memory tools, while API or SDK integration is recommended for guaranteed per-turn capture or recall.
- •UML claims features including a visual graph interface, updateable memory with superseded history, receipts for every save, backend-enforced saving rules, multi-source ingestion, and privacy protections such as no ads, trackers, data sales, or training on user data.