Documentation

BlogTracker is an AI-powered platform for monitoring and analyzing blog content at scale. You build trackers by adding blogs or posts, and the platform automatically surfaces clusters, narratives, sentiment trends, and entity networks so you can focus on understanding the content — not collecting it.

BlogTracker starts collecting and processing content automatically in the background. Analysis results fill in progressively — you will see clusters, narratives, and sentiment data appear as processing completes. You do not need to stay on the page while it runs.

BlogTracker includes: Cluster Analysis (groups posts by topic), Narrative Analysis (detects recurring storylines), Sentiment (tone over time), Posting Frequency (activity trends), Topic Distribution, Blog and Blogger Distribution, Morality Assessment, Word Cloud, and Influence scoring. All are accessible from a tracker's Deep Analysis section.

BlogTracker stores the post text, source URL, publish date, and author information where available. It also attaches analysis results — sentiment score, toxicity score, grievance score, detected entities, topic cluster assignment, and narrative membership — to each post for use in dashboard views.

There are two ways. From the Tracker page, open a tracker then use Add Blog to paste a blog URL directly. From the Explore page, enable Selection Mode, select one or more blogs or posts, then click Add to Tracker at the bottom — you can pick an existing tracker or create a new one on the spot.

The Trackers page marks a tracker Active while a crawl, topic analysis, or clustering run is working on it, and Queued when jobs are waiting. The Activity page lists those jobs with their progress, and you can ask the assistant "what is running?" or "is my crawl done?" from anywhere. A Deep Analysis section that is still empty after its job finishes may need re-triggering from that section's controls.

Yes. Go to the Trackers page and click the Compare icon, or navigate to /tracker/compare. Select a left and right tracker from the dropdowns — the comparison loads automatically and shows shared posts, blogs, entities, narratives, and clusters side by side.

Yes. Click the bookmark icon on any chart or insight card to add it to a report. You can name and organize reports from the Reports page, then export as PDF to share with your team. Reports capture a snapshot of charts without requiring you to rerun analysis.

Alerts are configured per tracker. Click the bell icon next to any tracker on the Trackers page to open the Alerts dialog, then turn on the tracker events you want to receive. Live alert events include new posts collected, cluster analysis re-run, and topic analysis re-run; planned event types such as toxicity spikes, grievance spikes, emerging keywords, new influential blogs, and narrative changes may appear in the UI as coming soon. When an enabled event occurs, the alert appears in the bell menu in the top navigation bar.

First, confirm your tracker has active sources and recent posts. Refresh the analysis page after a short wait since processing runs in the background. If results still do not appear after a few minutes, try re-triggering analysis from the relevant Deep Analysis section. If the issue persists, contact your admin with your tracker name and the time the issue started.

The chat bubble in the corner answers questions about the tracker you have open — every deep-dive chart, the topics and narratives, sentiment and grievances, images and videos, and cross-cutting questions no single page answers, such as "brief me on this tracker", "is anything unusual?", or "tell me about this person". It also explains how the platform works and can tell you which trackers you have, what a report contains, and which jobs are running. It knows which page you are on, so "explain this chart" works without naming it.

From your own data, not the open web. Every figure comes from the same tables and endpoints the dashboards use, so a number in chat and the same number on a chart have one source, and the assistant names the metric it used. It also states what it could not cover: analyses such as sentiment or grievance only reach the posts they have processed, so a percentage is a share of the analysed posts and the assistant says so. If you want to check a figure, ask it to show the posts behind it — you get the actual articles with links.

Almost nothing. Every analysis question is read-only. The one exception is saving a note into a report: say "save a note to my report: ..." and it writes the finding into a report you own, creating one if you name a new title, then quotes back what it saved. That works only for the tracker's owner, so anyone you share a tracker with can ask questions but cannot write. The assistant cannot start crawls or analysis runs — those stay on their own pages.