Lab, a personal project
The morning briefing
The pattern behind the morning brief: a scheduled pipeline that researches the news with an LLM and emails a short HTML brief every morning.
- Services
- Automation
- Stack
- An LLM, A cloud machine, HTML email
- Updated
Context
This one is a personal project, and it shows the pattern behind the morning brief: let a model do the reading, and send people the short version.
It answers one question before the day starts: what happened that is worth knowing about?
The AI night shift is the same pattern built for a business: it works through the business's own systems overnight and sends the owner a short brief before they open.
Features
What it does
A model reads the sources and follows up on the stories worth following, instead of stopping at the headline.
A second pass cuts the research down to a short brief.
The brief arrives as an HTML email every morning.
The machine stays stopped until its timer starts it, so it costs money only while it works.
Problem
News arrives all day from too many places, and reading it well takes time most mornings do not have.
A plain feed of headlines does not solve it. The useful part is the research: what a story means, and what else it connects to.
Approach
The pipeline runs on a schedule. It fetches the sources, runs an agentic research pass with an LLM that follows up on the stories worth following, then an editorial pass that cuts the result down to a brief.
The brief goes out as an HTML email every morning.
It runs on a cloud machine that stays stopped by default and starts on a timer, so it costs money only while it works.
Outcome
The brief is scheduled for every morning, and the machine sleeps the rest of the day.
Stack
- An LLMthe research and editorial passes
- A cloud machinestopped by default, started on a timer
- HTML emailthe brief itself
Tell us what you are working on.
A few lines about your business is enough to start. We reply with questions and a plain next step.