Personalization

Recommendations

Structured audience metadata plus declared profiles produce adaptive ordering — every recommendation carries an inspectable one-sentence justification. No surveillance required.

Choose your reader profile

Profiles are declared, not inferred — you tell the platform who you are; nothing tracks you.

Plain-language overview

Ranking content for your profile…

How ranking works

1. Declared match (+3): editors tag content for specific audiences at authoring time.
2. Interest affinity (+2): category overlap with followed topics when provided.
3. Recency (0–1): newer work ranks slightly higher within equal relevance.

Deliberately simple and auditable by design — see the personalization article for the reasoning. Adapted summaries use controlled AI transformations, labelled as generated.

Read the design rationale