Membase

Today, I recorded a podcast with the founders of @mem_base. They built a service that can set up a memory system for AI agents something that took me over a month to build in just one minute. 😱 @JoshuaIPark Cool.

Switched to Claude as my main AI 2 months ago but still use ChatGPT, Codex, and Perplexity daily. The problem: rebuilding the same context and projects across every tool. Every new chat starts from zero. That changed when I found @mem_base. It's a persistent memory layer that works cross-agent — Claude, Claude Code, ChatGPT, Codex, Perplexity, and more all share the same memory and knowledge base. It has two components: Memory — personal context that carries across every conversation Wiki — your own personal knowledge base (the concept Karpathy has been pushing, but actually integrated with your AI) Claude's native memory is one page of text for the whole account. Projects have separate memory so they can't reference each other. Membase fixes all of that and unlocks a level of AI memory I didn't even know was possible. Get 1 month of Pro free 👇 app.membase.so/r/B84T4DYR

Excited to join Hermes Agent Meetup @ Seoul as a speaker this Saturday! I’ll be sharing how memory can make your Hermes agent truly understand you: your context, preferences, and workflows. See you there 🚀

Interview Series — Episode 1 How do founders use @mem_base? One of our users, Roy Tan (former Head of Public Policy at Meta), now runs an AI marketing consultancy helping SMEs adopt AI workflows. He works across a PC, Mac mini, and laptop, switching between Claude, ChatGPT, Codex, and local models like Llama to build CRM-style tools, content automation, and client workflows. Before Membase, he had to manually move markdown files between devices or stay locked into a single machine. Now, he saves project context to Membase on one device or model and recalls it on another, keeping his workflows moving without losing context. For consultants and builders working across devices and multiple LLMs, @mem_base acts as a shared memory layer that makes workflows portable across machines and models. Try it free: membase.so

Heremes Agent is now available as a native Membase plugin. Install it with a single command and give your shared, persistent memory and wiki to your Hermes Agent.|

@mem_base Discord just crossed 800 members!! Thank you so much for joining our community, even if it's in the early beta. I'll make this reach 1M.

The Membase plugin for Cursor is officially live! Get persistent, self-evolving memory for Cursor and know your preferences, todos, failures, and successes.

Link: cursor.directory/plugins/membas…

We just updated the Wiki on @mem_base, inspired by @karpathy's LLM-Wiki and @garrytan's GBrain. Thanks to Garry's repost, my previous post went viral, and we saw a huge spike in sign-ups. So we decided to offer a 1,000 Pro plan coupon in his honor: GARRYTAN We're so excited to see how people will use this amazing second brain. Be quick. The coupon is valid only this week.

Amazing day. We got 200 users in half a day after @garrytan reposted my tweet. Tomorrow we'll launch the next version of @mem_base with Wiki. Thinking of giving a free Pro plan for my followers. Stay Tuned!

I'm glad people are discovering that @mem_base is the most seamless 'second brain' in the world.

@mem_base just crossed 500 users without paid marketing, just word of mouth. 4 months of building a personal second brain that's: > Portable - not locked into any single platform > Easy - no prior knowledge needed to get started > Agent Friendly - plug into any agent you already use Under the hood: Graph/Vector RAG, RRF reranking, embedding pipelines, token optimization. On the surface: it just works seamlessly Just getting started.

Huge updates are coming soon in @mem_base. We made @karpathy's LLM-Wiki + @garrytan's Gbrain into a real working product. Memory, Wiki, auto-update, and live sync with agents. Anyone can get their own second brain with a few clicks.