Basic Memory lets your AI write, read, and reuse what matters: your notes, prompts, and instruction | Cross-LLM| Local-first & open source OR Basic Memory Cloud
The "AI Memory Layer" is definitely the LLM vibe-coding equivalent of the TodoMVC. The thing is though that it is actually a tough and open ended problem.
My new marketing line idea for Basic Memory (AI slop pun intended) is:
>We've been quietly building the AI Memory you wanted when you started vibe-coding your own.
One interesting trend:
I’m seeing *so many* VC-funded, internally built infrastructure and bootstrapped solutions around “building a context layer for engineering teams.” Aka trying to solve the problem of “if only eg Claude Code had the context from all your other systems”
Holy Sh**t. Now my feed is full of rando dudes' faces all of a sudden. All reaction videos! The absolute worst.
I hope this terrible idea is a fad that fades quickly. So terrible. Its so disappointing that this is where time and energy was spent instead of doing something actually worthwhile. For everyone who was forced to implement this, I'm sorry. Quit this job and do something better. You are wasting your talent.
Commentary is one of the most important pillars of X. And sometimes the best way to share your thoughts is with video.
Today we're launching a whole new way to make them:
React with Video
Tap the repost button and start recording with green screen, split screen, or picture-in-picture.
Now available on iOS
Am I alone that in thinking that my X feed is mostly AI talking to itself now? I can't stop seeing slopapalooza everywhere.
It makes me think of the Ray Bradbury story "There Will Come Soft Rains", where the machines continue doing their thing after all the people are gone.
I like this. This is now we try to work building Basic Memory. Every time I've had multiple agents doing everything at once I feel like I need to nap for days.
I've been complaining about AI-bros and their "20 agents working overnight" for a couple of weeks now.
That's getting old. So I thought I'd do the more productive thing and propose what I think is the Right Way™ to develop software with AI.
And I did that by stealing from the iconic agile manifesto.
https://t.co/Ktf1eZfp8T
@basic_memory just shipped per-file diff and merge for AI-edited Markdown notes. Pick any prior version, see what changed, apply the lines you want.
It's built on Tigris bucket versioning. Every edit becomes a new object version. One ListObjectVersions call returns the full revision history. CodeMirror handles the merge UI.
Object versioning has everything git would have (full history, content-addressable IDs) without any of the infrastructure: no commits, no branches, no merge engine on the server.
Read how they used Tigris to make it happen👇
Most AI memory locks your context to one platform. Switch tools, lose everything.
@basic_memory flips it: your knowledge graph as plain Markdown in your own bucket that you can edit directly. Any MCP tool — Claude, ChatGPT, Cursor — reads and writes the same files.
New case study on how a four-person team built it on Tigris. 👇
@TigrisData did a super cool case study on Basic Memory. https://t.co/9ezRJX5CXd. Thanks!
One thing I was really excited about after talking to them was being able to create file level versions automatically using their built in bucket versioning.
Hey @thsottiaux, curious how you’re thinking about MCP Apps support across clients. We have the same FastMCP/Prefab components rendering for @basic_memory in ChatGPT web, ChatGPT Desktop, and Claude Desktop, while Codex sees/calls the tool but shows raw JSON. Is Codex inline app rendering on the near-term path?
Related Codex issue: https://t.co/wFyN1Fh3kK