We’ve been experimenting with what a real *“always-on” proactive agent* could look like, and built one called *memUbot*. It runs 24/7 in the background, keeps track of what you do, and builds long-term memory from it. Over time, it learns your habits and context. Instead of waiting for prompts, it tries to infer your intent and act before you ask. The goal is a more “assistant-like” proactive agent: - observe - remember - act - continuously One practical detail: it’s designed to be cheap to run. By keeping context in memory and minimizing LLM calls, it uses far fewer tokens than bots like moltbot. You can try it here (download & run locally): https://memu.bot/ If you’re already using moltbot and don’t want to migrate, you can plug in our open-source memory layer, *memU*, to upgrade its memory and make it truly 24/7 and proactive: https://github.com/NevaMind-AI/memU Happy to hear thoughts from people building agents or infra in this space. |