Skip to main content

Our Methodology

How AI helps us create the Daily Sense-Making Brief

1 Trusted Source Gathering

We continuously monitor 140+ carefully selected news sources, including quality journalism outlets, policy publications, think tanks, investigative journalism, and intellectual podcasts. Each source is rated for political leaning using AllSides methodology.

News & Analysis Sources

Left

The Guardian The New Yorker The Atlantic The Independent The Intercept New Statesman Freddie deBoer Matt Taibbi

Centre-Left

ProPublica Slow Boring Noahpinion Commonweal

Centre

BBC News Financial Times The Economist Bloomberg Axios Foreign Affairs Foreign Policy Politico EU Stratechery Semafor Rest of World The Conversation Al Jazeera English Lawfare MIT Tech Review Carbon Brief SCMP Ars Technica War on the Rocks TechCrunch Farnam Street Brookings

Centre-Right

UnHerd The Dispatch The Free Press Reason The Critic Marginal Revolution Cato Institute Andrew Sullivan

Right

The Telegraph The Spectator National Review The American Conservative City Journal Manhattan Institute Commentary First Things Christianity Today

Podcasts

The News Agents The Ezra Klein Show The Rest Is Politics Lex Fridman Podcast Diary of a CEO Modern Wisdom The Tim Ferriss Show Huberman Lab Acquired Podcast All-In Podcast Triggernometry

Political leanings are based on AllSides ratings where available, with manual assessment for podcasts and newer publications. We update ratings monthly and add new sources regularly.

2 AI-Powered Relevance Scoring

Each article is analysed by AI to assess two key metrics:

  • Civic Relevance (0-100%)

    How valuable is this for public understanding? We prioritise policy, institutions, rights, and long-term societal impact over celebrity news and product reviews.

  • Clickbait Risk (Low/Medium/High)

    We filter out sensationalised content. Articles with high clickbait scores are deprioritised.

3 Intelligent Topic Clustering

Using AI embeddings, we group related articles from different sources into unified topics. This means when multiple outlets cover the same story, they're clustered together so you can see how different sources frame the same event.

Why this matters: Rather than reading the same story 5 times from 5 outlets, you see one consolidated story with multi-source coverage analysis.

4 Coverage Balance Analysis

For each story, we show which outlets are covering it and from what political perspective:

Example coverage distribution
Left-leaning Centre Right-leaning

Source ratings come from AllSides Media Bias Ratings. We show the balance not to say who's "right," but to help you see blind spots in coverage.

5 AI-Generated Summaries & Perspectives

For each selected story, AI generates:

  • Key bullet points - The essential facts in 3-4 sentences
  • "Why This Matters To You" - Personal relevance and impact
  • Left/Centre/Right perspectives - How different viewpoints frame the story
  • Blindspot analysis - When coverage is skewed, we explain what might be missing

6 Automated Daily Publication

The Daily Brief is generated and published automatically each evening at 6pm UK time. This allows for same-day coverage of developing stories while ensuring you have time to digest the news before the next day.

Transparency note: This is an AI-first product. While we've designed the system to prioritise accuracy and balance, AI can make mistakes. We always link to original sources so you can verify for yourself.

What We Don't Do

  • Sensationalise headlines for clicks
  • Tell you what to think
  • Pretend to be objective (we show you the perspectives instead)
  • Cover everything (we focus on what matters for civic sense-making)
  • Hide our AI use (we're transparent about how this works)

Transparency

Society Speaks is built with transparency in mind. Our code is publicly available under the AGPL-3.0 license, which means you can inspect exactly how our AI systems work and suggest improvements. We believe civic technology should be accountable.

View Source Code