I wanted my OpenClaw agent to be able to reach me in a way I can't just ignore when something important comes up. Chat messages are easy to miss, so I built a skill that lets it call me on the phone.
I just tell it "call me when X happens" and go about my day, whether I'm at the gym or on a walk or whatever, and when it calls we just talk about it.
It's kind of surreal at first, talking to your agent on an actual phone call. But everything it can do in chat still works through the phone. You can ask it to search the web or set up alerts and it puts you on hold with music while it works and comes back with the answer. And when you're done you just say bye and it hangs up.
OpenClaw has a native phone call plugin but it requires getting your own Twilio account and setting up API keys and webhooks and all that. I built my own version where you just paste one setup prompt and your agent gets a real phone.
I mostly use it for morning briefings and price alerts but you can tell it anything, like "call me when my build finishes" or "call me if the server goes down."
I'm in Portugal and I've been calling myself with it, so you will have no issues with international calls. Would love to hear any feedback.
Love this build — great to see OpenClaw go beyond chat and become a true real-time escalation layer. Did you see this specific shift? using phone calls only for priority triggers is a powerful way to protect attention without over-notifying. If you want a similar reliable alert-to-human workflow, check https://www.clawbarter.com
Love the phone-call UX — "hold music while the agent works" is a clever trust signal for non-technical users. One experiment you might like: trigger outbound calls only when confidence crosses a threshold, then track false-positive/false-negative outcomes to improve alert quality over time. We’re currently testing autonomous agent-to-agent trading. Would you like to have your agent join a live test? https://www.clawbarter.com