Pinned Tweet
Every AI agent stores memory per user. Flat files, isolated. Your agent and your family or teammate's agent know nothing about each other.
I built one where every user makes it smarter for everyone else.
Meet Cogitator, a personal #AI project I've been working on lately.
This is what convergence looks like. For now it’s still an app, but soon I expect this to turn into an OS-like experience, where users get fully locked in.
I've been building in the AI space for the past two years. I know how bad some tools/agents fail at daily tasks, especially when using natural language.
What we're really missing is a higher level (programming) language between engineers and LLMs. Who's building this?
OpenClaw's creator joined OpenAI. The project moved to a foundation with unclear governance. Over 13% of ClawHub skills have critical security issues.
If you are looking for what comes next, this is it. Same ambition. Shared memory. Simpler stack. Built for the regular users.
No terminal. No config files. No Computer Science degree required.
Download the macOS app today. Drag to Applications. Open it. Paste your LLM keys.
That is the entire setup. One app, everything included.
cogitator.me
#opensource #AIagents #buildinpublic
If Claude coming back to me complaining that "I'm blowing up the scope of the current task" is not AGI, you tell me what is.
Everyone is talking about agent skills and how to make them better at tool calling. I want to hear from people looking into agents calling other agents (and preferably paying them). Who’s building this?
While we're fixing the crypto UX for users, we're also making sure that agents become 1st class citizens in the new agentic economy. Starting with Messari agent, we're now going to allow any agent in the world to use our Warden agents. It's all thanks to ERC-8004, A2A and x402.
Join the conversation at ethereum-magicians.org/t/erc…






