Product Update

Pair Intelligence: AI Analysis of Bilateral Country Relationships

April 18, 2026  ·  Global Perspectives

Country briefings tell you what is happening inside a country. But most of the world's most consequential dynamics happen between countries — in the space where two states' interests, histories, and power calculations collide.

That is what Pair Intelligence is built to analyze.

What it is

Pair Intelligence generates a structured bilateral analysis for two countries — drawing on 30 days of live news coverage, daily country briefings, narrative thread analyses, and an operator-verified editorial fact layer. It is updated continuously and covers 15 country pairs across the most active geopolitical relationships today.

Each analysis contains seven sections:

• Current State — a narrative account of where the relationship stands right now, grounded in recent events
• Root Driver — a three-layer analysis: the immediate trigger, the structural pattern behind it, and the deep historical context that made it possible
• Timeline — a chronological sequence of the 5–7 most significant events shaping the current situation
• Trajectory — three plausible scenarios for how the relationship develops over the next 30–90 days
• Predictions — specific, time-bounded claims with confidence levels (High / Medium / Low) and the causal mechanism behind each
• Watch Items — specific actors, indicators, and signals that will tell you how the situation is actually evolving
• Data quality disclosure — a transparent rating of how much source material the analysis is based on

The 15 pairs live today

The initial set is drawn from the most co-occurring country pairs in 30 days of Global Perspectives coverage — the relationships that keep appearing together in the same stories.

Each pair is classified by data quality based on how much recent coverage exists:

Pairs with sparse or limited data show a disclaimer. The analysis is still generated — but you know exactly how much source material it rests on.

How the analysis is grounded

Pair Intelligence does not rely on a language model's training data alone. Each analysis is built from a structured evidence stack:

• Editorial fact layer — operator-verified facts about current leadership, conflict trigger dates, and government transitions. These override everything else. This is how the Iran × Israel analysis correctly identifies Mojtaba Khamenei as Supreme Leader following the February 28 airstrike — a fact that live news search alone consistently missed.
• Automated leadership data — daily-updated head-of-state and head-of-government records sourced from Wikidata's structured political data, cached in our database. Updated every 24 hours.
• 30-day archive — deduplicated event entries from the live news pipeline, filtered for the pair's countries and ranked by relevance
• Country intelligence profiles — each country's daily briefing, including trajectory and risk signals, feeds into the pair analysis
• Thread analyses — narrative arcs from the thread intelligence system provide the story-level context that raw articles lack

Why bilateral analysis is different from single-country briefings

A country briefing answers: what is happening inside Iran right now? A pair analysis answers: what is the dynamic between Iran and Israel, and why does it exist?

These are different questions. The bilateral frame forces the analysis to do something a single-country view cannot: it has to account for the interaction — the feedback loops, the asymmetries, the points where each side's calculus is shaped by the other's actions.

The Root Driver section is where this is most visible. Layer 1 is the immediate trigger — a specific event, actor, and date. Layer 2 is the structural pattern that made that trigger consequential. Layer 3 is the historical context that made the structural pattern possible. Reading all three together is a different kind of understanding than any headline provides.

What comes next

The current 15 pairs are a starting set. We are expanding to cover more relationships as coverage volume supports it — European pairs, Southeast Asian dynamics, and African relationships where the news pipeline generates enough material for a grounded analysis.

We are also evaluating ACLED conflict event data as a structured supplement to the live news pipeline — which would significantly improve the depth of analysis for active conflict pairs.

Cross-links from Country Intelligence pages to relevant pairs are coming shortly — so the Iran country briefing will surface the Iran × Israel and Iran × United States pair analyses directly.