A newsletter
for one.
A daily briefing on AI and technology, made for my own curiosity.
- FetchRSS feeds
- ReadArticle text
- GroupRelated coverage
- CurateAI selection
- DeliverDaily email
- RefineFeed feedback
↶ Feed feedback shapes the next briefing.
What it is.
This project collects stories from RSS feeds, groups overlapping coverage, and uses AI to curate a daily email briefing. The feed list evolves over time through a feedback loop. It currently serves one reader: me.
The aim is a useful overview, with a small selection of stories and enough context to help me decide what to read next.
Why I made it
I want early signal.
Technology moves quickly, and some things reach my radar later than I’d like. I wanted a way to catch useful signals earlier and keep up with what’s happening in AI and technology.
The idea is to give me something useful even on a day when I only have a minute, while leaving a way into the stories I want to spend more time with.
Follow the process.
From the sources to my inbox, then back around to improve the sources. Open a step to look a little closer.
Collect the stories.
Pull recent stories from a configurable set of RSS feeds. The workflow looks back over the previous 36 hours.
Get beyond the headline.
Fetch the full article text so the system has more context for comparing stories and spotting coverage of the same event.
Find the overlap.
Group near-duplicate stories and choose a representative for each event before asking the model to curate the briefing.
Choose what’s worth a look.
Send those representatives to Claude to select six to eight items, write concise summaries, add signal tags, and create an overview of the news.
Bring it to my inbox.
Format the briefing as an HTML email and deliver it through the automated workflow. The documented schedule is 7 AM Pacific each day.
Keep improving the sources.
Track how feeds perform over time, put weaker sources on probation or disable them, and recover those that improve. A weekly discovery process also looks for new feeds and ranks recommendations. That feedback helps shape the sources for future briefings.
Under the hood.
The code, feed configuration, and workflow documentation live in the public repository.