Hey @gabbyesposito, just following up on this:
For search, results are ranked using a custom ranking system (not end-user configurable) that prioritises content in this order: articles, federated content, answered posts, and then last updated date. Quality signals such as likes, post count, and views are also factored in. There's no way for members to manually sort or filter results, and ranking settings are managed at the index level only. So while recency is part of the mix, it won't always surface as the top signal depending on the other ranking factors. That's likely why adjusting the AI agent context for recency led to some unexpected results!
For Related Topics, those are generated automatically by the search engine using the current article's title as the query. Content with overlapping title keywords is more likely to surface. There's no manual curation, per-article pinning, or personalisation available, so every visitor sees the same related content (scoped to their permissions).
Hope that helps clarify! If you have ideas for how you'd want these to work differently, feel free to drop them in the Ideas section.
