In this video, I sit down with Alex Finn and give what I think is the clearest explanation of Clawdbot/OpenClaw (now officially renamed Moltbot) and how to use it like a real 24/7 AI employee—not a Tamagotchi. The big idea: treat the agent like a proactive teammate. Alex calls his “Henry,” runs the relationship through Telegram, and wakes up to a daily morning brief. While he sleeps, Henry monitors trends, researches ideas, and actually ships work as pull requests for review. We break down what makes this work: deep context + explicit expectations for proactivity. You don’t just turn it on and magically get a SaaS built. You have to feed it everything about your business and life, then set the rule: “do work you think improves my business and makes me money—ship PRs, don’t push to prod.” We also talk model strategy (Opus as the brain, Codex as the muscle for cost control), the “Mission Control” Kanban board Henry built to track autonomous work, hardware/hosting tradeoffs (cloud VPS vs local Mac Mini/Studio), and the security mindset—prompt injection is real, so don’t hand it the nuclear codes.