WORLDINTELLIGENCE
.live · global situation
UTC--:--:--
CRITICALEastern EuropeSustained energy-infrastructure strike pattern observedHIGHRed SeaVessel-traffic anomaly continues in southern transit laneHIGHEast AsiaMultiple grey-zone incursions across median line in 24h windowMEDIUMSahelCivilian-protection deterioration in northern provincesMEDIUMAndesMagnitude 6.1 seismic event, shallow depthLOWGlobalCrypto OTC desks observe elevated cross-border settlement volumesCRITICALEastern EuropeSustained energy-infrastructure strike pattern observedHIGHRed SeaVessel-traffic anomaly continues in southern transit laneHIGHEast AsiaMultiple grey-zone incursions across median line in 24h windowMEDIUMSahelCivilian-protection deterioration in northern provincesMEDIUMAndesMagnitude 6.1 seismic event, shallow depthLOWGlobalCrypto OTC desks observe elevated cross-border settlement volumes
How to use WorldIntelligence

User guide

A 5-minute orientation to every section of the site. Where it says AI, that feature is powered by a large language model — read the AI section below for what that means in practice.

AI features at a glance

Four places on the site are AI-powered. Anywhere you see the AI badge, a language model is involved. Anything not badged is straight data — no AI in the loop.

Important: AI output is decision-support, not decision-authority. We do not store the contents of your questions on our servers. The AI is grounded only on the data shown on the site at the time of the request — it does not browse the open web during a reply.

Tour the rest

How our AI works

  1. Grounding. Every AI call sends a structured snapshot of the current state of the site — top headlines, country risk index, active alerts, market levels — into the prompt.
  2. Constraints. The model is instructed to stay factual, neutral, short, and to refuse speculation outside the snapshot.
  3. Provider chain. Requests run through a free-tier model first, with automatic fallback to a paid frontier model if upstream is rate-limited or unavailable.
  4. Caching. The hourly World Brief is cached on disk to avoid spending quota on every page load. The Analyst chat is not cached — each reply is fresh.
  5. Markdown. If the AI returns a markdown table or list, we render it as proper HTML — no raw asterisks in the UI.

Keyboard tips

Open AI Analystclick bottom-right button
Pan mapdrag
Zoom mapscroll wheel / pinch
Fullscreen video⤢ icon on each panel
Close modalclick outside / Esc