← All help topics

Trends

The “Trends” section surfaces hot topics: AI analyses fresh material from your sources and finds real events, facts and announcements. You can prepare a post on a trend right away.

1🔥 Turning trends on and off

At the top of the section there's an “Analyse hot trends” checkbox. Trends are background AI requests and they spend tokens.

  • ✅ On (default) — the list is computed on schedule and via the “Refresh trends” button.
  • ⛔ Off — computation stops, no tokens are spent on it (neither on schedule nor via the button).

Need them again — just tick the checkbox back on.

2What counts as a trend

The service collects fresh material (titles and snippets) from active sources in «Sources» and passes it to the model. It extracts REAL topics — concrete events, facts, announcements, significant news, not just frequent words.

The MORE sources a topic surfaces in, the “hotter” the trend (shown on the card: number of sources). A real trend is a topic covered by several sources at once (at least 3 for auto-posting) — that's what signals its importance. Topics with one or two sources may also appear in the list as “emerging”, but the service auto-publishes only genuinely multi-source ones (and gives them a deeper analysis checked against scientific bases); a single-source one can be posted manually.

3What trends need to work

Trends are computed only for a filled and active project. You need:

  1. a filled “Marketing context” — the AI uses it to judge topic relevance (without context trends aren't computed);
  2. a configured model in «AI providers & API keys»;
  3. at least one active source in «Sources»;
  4. the project not paused.

The more on-topic the sources, the more accurate the result.

4The trend card

Each trend card shows the essentials, so you can quickly decide whether to react:

  • 📊 frequency — total mentions;
  • 📣 number of sources where the topic appears;
  • 🔗 links to the specific material;
  • 🗓️ time it was formed (“Updated …”) — how fresh it is.

5Make a post from a trend

A trend card has two buttons:

  • ✍️ “Make a post” — generates a draft on the trend's topic (using the project context, language and region) and places it into «Plan & post queue» for approval; the trend is marked as used.
  • “Hide” — removes an uninteresting topic from the list.

Generation needs a configured text-model key in «AI providers & API keys».

🤖 Auto-posting from trends (if plan auto-approval is on) is stricter than the manual button: a post is created only for a “real” trend — a topic surfacing across SEVERAL sources at once (at least 3 by default) — and only if a post with the same meaning has NOT been published yet (including as a regular source post), so the same event doesn't go out twice.

🔬 A trend post is not a rephrase of one item but an ANALYSIS: the service synthesises material from several sources, additionally checks the topic against scientific bases (prior research, objections, important caveats — with links in a “Sources” block) and plainly notes the topic is drawing wide attention. So “a lot of buzz around a topic” becomes one thoughtful, verified post rather than several shallow ones.

A single-source topic isn't treated as a trend and isn't auto-published, but you can still post it manually via “Make a post” (the analysis and enrichment apply to it too).

6“Trends” and “Archive” tabs

The list is split into two tabs (like in «Sources»):

  • 🔥 Trends — current hot topics you can still post about.
  • 🗄️ Archive — topics you've already posted on, plus stale trends (older than 3 days / off the agenda).

This keeps the active list uncluttered while the history stays available separately. A topic you've posted on moves to the archive automatically.

7Refresh: manual and automatic (economical)

The list refreshes in two ways:

  • 🔄 “Refresh trends” — recompute right now (a configured model is required).
  • ⏱️ Automatically in the background — at most about twice a day.

The background run saves tokens and never calls the model in vain:

  • if the set of source material hasn't changed since last time, the analysis is skipped (the same information isn't sent to the model again);
  • 🔒 recomputation pauses during the project's quiet hours and for paused projects.

So trends spend almost no tokens when there's no news.

Didn't find an answer? Write to us — we'll help.

Contact support