Make a daily audio briefing from your newsletters

They had an agent read selected AI newsletters overnight and turn the highlights into a podcast episode ready for the morning.

Done with Custom agent· Intermediate· Oct 2026

What they did
On a Mac Mini in Claude Cowork, they linked the agent to Gmail and asked it to create a 5 AM scheduled run. The job checks a chosen mail label, skips about 18 lower-value newsletters, reads the rest, and drafts a recap. A local Qwen3-TTS model turns that text into MP3 audio, and the finished file is published on GitHub Pages with an RSS feed; they picked the voice and audio setup.
How it went
They spent about two hours getting it working in one session. By the next morning, the first episode, about 8 to 10 minutes long, was already showing up in Apple Podcasts. The voice configuration was the messiest part: the macOS voice could not be used, and edge-tts was unreliable.
Tips
Local TTS can be the slowest and most fragile part of the build: Qwen3-TTS took roughly 30 minutes to install, hit dependency conflicts, and got stuck on a download. It also climbed to 6.5GB of memory on a 16GB Mac, and they were using the Claude Max plan at $100/month.

Try it yourself with OpenClaw

Every night at [time], look through the messages in my [Gmail label or folder] from the last 24 hours, identify the newsletters worth keeping, and write a conversational [length]-minute script for a morning listen. Stop after preparing the draft and ask me to approve it before you do anything final. Once I say yes, generate the audio, place it in [destination], and send me the file location plus a brief rundown of what it includes.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on bronsonelliott.substack.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.