About

The Brief

What The Brief is, how an AI agent assembles it each morning, and how it chooses stories.

The Brief is a daily digest of about ten articles published since the previous brief. An AI agent selects and summarizes the articles each morning. Nicolas Moutschen maintains the site as a personal reading list and publishes it for anyone who shares the same interests.

Scope

Each brief crosses three divisions:

  • Engineering: software, systems, architecture, security, AI/ML tooling, and programming languages.
  • Social sciences: sociotechnical systems and safety, organizations and management, economics, psychology, and sociology.
  • Humanities: philosophy, ethics, ideas, and culture.

A typical brief holds five engineering stories, three social-sciences stories, and two humanities stories. Non-engineering feeds publish less often, so engineering fills the gap on thin days. Each brief keeps at least one social-sciences story and one humanities story when a suitable candidate exists. The selection favours topics relevant to a principal engineer who works on critical system safety and sociotechnical systems.

How a brief is made

  1. A scheduled job runs every morning at 07:03 CET and reads a curated list of RSS and Atom feeds.
  2. The agent drops stories that a brief from the last seven days already covers.
  3. The agent ranks the remaining candidates by news value, division balance, source diversity, and the reader preferences below.
  4. A separate extraction agent reads each selected article and writes its summary. The brief copies that summary without edits.
  5. The agent records a verdict and a reason for every candidate, kept or rejected, in an audit trail.
  6. An automated check confirms that the audit trail matches the published stories before the site builds.

Reader preferences

The site owner votes on stories they like or dislike. Once a week, a job condenses those votes into a short preference profile and weights recent votes more heavily. The agent treats the profile as a soft signal within each division. The profile never overrides the division mix or news value.

Reviews

Reviews synthesize several months of briefs into one essay, with charts of the division mix over the period.

Caveats

An AI model writes every summary. Summaries can omit context or misstate details. Follow the link to the original article before you rely on a claim.