AI automation demo
You are viewing: News
SignalOS turns news into a clear decision, with proof.
Live mode checks the latest saved data.
SignalOS turns news into a clear decision, with proof.
A simple log of what the app has read.
The app collects relevant Iran and U.S. updates.
It decides if the news matters for the prediction.
Important news can move the app prediction.
Every read item stays visible here.
Saved source notes stay here even if a live feed removes or reshuffles old stories.
Live feeds can drop old items. The archive keeps copied NYT, FT, IAEA, WSJ, and other source notes visible for the demo.
Feed is temporary. Archive is the permanent research record.
Outlet, URL, original headline, publisher time, and ingestion time.
Claims, actors, event type, linked market, and source confidence.
Features, model version, SignalOS estimate, Polymarket probability, and timestamp.
Final resolution, Brier score, log loss, and calibration bucket.
The professional sources this demo is designed to monitor or archive.
How the app treats major sources without flattening their language into one generic summary.
Useful for broad conflict framing, official reactions, and public-facing political context.
Useful for sanctions, assets, energy, shipping, and negotiation mechanics.
Useful for first-pass event confirmation before comparing analysis-heavy sources.
A simple keyword pass over the archived news so we can compare whether the update logic makes sense.
This is a lightweight archive check, not the final production model. It lets us compare whether the saved news points in the same direction as the 94% outside view.
How the archived news moved the June 30 YES read from March 18 onward.
Newest first. Evaluated items show whether they changed the prediction.