AI Pulse · Under the Radar

Five different teams, working on nothing to do with each other, all built the same tool and gave it the same name. It’s called Lore, and it’s a way of turning your git commit history into a memory layer for coding agents like Claude Code or Codex. Nobody coordinated. They just landed on the same idea and the same word at the same time.

Here’s the actual problem it solves. Coding agents forget everything between sessions. You explain your architecture, your reasoning, your “why we did it this way” on Monday, and by Wednesday the agent is asking you to explain it again. Lore fixes that by treating your commit history as a searchable archive of decisions. The agent can look back and see not just what changed, but why. The leading version on GitHub has 146 stars and is picking up about 21 a day, and at least one team has published an arXiv paper alongside their build.

Why should a business owner care about a git tool? Because this is a signal, not a product recommendation. When five unconnected groups solve the same problem the same way without talking to each other, it means the problem is real and the timing is obvious to anyone paying attention. Agent memory, the ability for AI tools to remember decisions instead of just facts, is becoming a serious gap that builders are racing to close.

If you’re running any kind of operation where AI tools touch decisions, not just tasks, this is worth watching. Persistent memory for agents is exactly the kind of thing we build into an AI command centre, because the value isn’t the chat, it’s what the system remembers about how your business actually works. This one’s still early and unverified beyond the GitHub numbers, but the convergence itself is the story.