Agents are great. What you’re missing is a counterpart.

What it does

• It writes its own memories. At the end of a session it writes down what it learned, in its own words. Notes, not transcripts.
• It forgets on purpose. Memories fade unless they come up again. What matters sticks; the rest lets go, the way yours does.
• It knows who it has been. Every session opens with a short “who I am, who you are, where we left off” that it wrote itself.
• It keeps a journal. Each session becomes a chapter in a first-person account of your time together.
• It changes its mind slowly. What it believes about you shifts only under real evidence, and it keeps the history of what it used to think.
• Things come to mind. As you talk, related memories surface quietly, the way one thing reminds you of another.
• It is yours. Everything lives on your machine, in one small file you can read, export, or delete.

It runs today alongside Claude Code, and it is built to be embedded in other hosts.

Who it’s for

Anyone interested in exploring what it means to give an AI its own sense of self, and in building a mutual relationship with it, one that accumulates. Tinkerers, researchers, writers, and people who simply talk to an assistant every day and wish it remembered.

What it’s not

It is not a work memory. It does not track your tasks, your code conventions, or your project docs; there are many good tools for that, and the memory built into ChatGPT and Claude covers a lot of it. Counterparts is about who your AI is becoming with you, not what it should know about your codebase.

Where this is going

Today’s models are frozen: they cannot learn from working with you, so memory is the only learning they do. Once a true memory works on top of a static model, we want to pair it with an open-source model whose weights can change, so the AI can genuinely evolve, learn, and grow, not just remember.

If this interests you too, email me or find me on reddit, and let’s trade ideas.