Two methods scan. One trend confirms.
You write what is actually worth writing.

Two independent research methods scan your industry. The weekly Deep Research method surfaces macro-statistic trends with citations. The daily Market Surveillance method catches specific near-real-time events. When both methods detect the same underlying trend, ILLIXIS clusters them into one entry and stamps it with a "Confirmed by N sources" badge. Multi-source clusters are stronger pillar candidates than single-source detections because two different research approaches agreed.

It is a two-stage process. First, an embedding similarity gate identifies candidate pairs. Pairs above the threshold pass to an AI pair classifier that makes a binary same-trend or different-trend call. The embedding stage filters out the vast majority of unrelated pairs before the AI runs, so we get accuracy without the cost of running the classifier over every possible pair.

It is a 0-100 score reflecting how strongly a trend warrants flagship content given your niche, not just its general newsworthiness. Scores of 80 and above are pillar candidates worth a long-form anchor piece. Scores of 50 to 79 are topic briefs worth a mid-depth article. Below 50 is spoke content, worth a supporting piece if the rest of the cluster fits. Trends are sorted on the dashboard by highest pillar potential first.

Generic trend tools use one signal, usually search volume or social mentions. Each source has its own bias. Search volume favors recurring queries. Social mentions favor outrage. ILLIXIS uses two research methods that look at different parts of the web. When they agree, you have signal. When they disagree, you have noise. Single-source tools cannot tell the difference.

Yes. Each trend supports a status workflow: new, reviewed, accepted, brief_created, dismissed, and aged_out. Clicking "Create Brief" seeds a content brief pre-loaded with the trend summary, suggested angle, citations, and data points. The research context flows directly into the brief, so nothing is lost in translation from discovery to writing.

Market Surveillance scans run daily at 02:30 UTC. Deep Research scans run weekly on Monday at 06:00 UTC. Cross-source clustering runs after each scan completes, plus an hourly safety-net job catches any tenants whose scans were tier-skipped. New trends typically appear by Monday morning each week.

Older trends auto-expire to the aged_out state when newer detections displace them via the weekly cap rotation. The dashboard stays focused on what is actually current. Manually dismissed trends stay out of the active queue without being deleted, so you keep an audit trail.