Mosaic News

Methodology

This is the plain-English guide to how Mosaic works. Each chapter explores how different parts and processes of how our app works for transparency and accountability

Chapter 01

How we use AI to write Story summaries.

See what the model reads, the rules it follows, and why every summary remains a starting point rather than a verdict.

Read this chapter

Chapter 02

How we use your Feed Formula to rank your feed.

Follow the five signals behind every score, see how your own controls change what rises, and where each number comes from.

Read this chapter

Chapter 03

Where your reading history lives.

See what is written down when you open an article, where those records stay, and how they become your private weekly reflection.

Read this chapter

Chapter 04

How we connect articles into Stories.

See how Mosaic compares meaning instead of matching words, so articles covering one event gather into a Story, and Stories into a Big Picture.

Read this chapter

Chapter 05

How we work out what each article is about.

See the labels attached to every article before you read it, the checks a topic has to pass before it is applied, and how the list of topics is kept current.

Read this chapter

Chapter 06

Where a publisher's ratings come from.

See how Media Bias/Fact Check rates an outlet's lean, factual record, and credibility, and how to check any of their work yourself.

Read this chapter

Chapter 07

Where the news comes from.

See which newsrooms Mosaic reads, the checks every arriving article passes, and why carrying an outlet is not endorsing it.

Read this chapter

Chapter 08

How Moods filter the whole feed.

See the one rule behind each mood in the menu, why a mood narrows the feed without re-ranking it, and how to build your own.

Read this chapter