Geopolitical analysis is moving from explaining events to forecasting their physical consequences. Clock&Cloud records analyst judgment continuously as structured "Clues," not just in periodic reports, so every forecast can be scored against what actually happens. Trust becomes a matter of measured performance, not reputation. AI enables this at scale, but only with deep domain expertise behind it.
By Colin Reed, Chief Intelligence Officer, Clock&Cloud
A tanker slows outside a strait it has transited a hundred times. A customs notice appears overnight on a ministry website. A port authority announces "technical delays" without elaboration. None of these are front-page events, and yet each of them has the capacity to completely upend global commerce.
For companies that operate across borders, the question is no longer whether geopolitics matters. It is how quickly and how reliably its consequences can be understood. This article sets out how geopolitical analysis is changing in the age of AI, why the traditional model of intelligence analysis can no longer keep pace, and what a more rigorous, measurable, and useful model looks like.
From "What Happened?" to "What Does This Mean for Us?"
Our discipline is changing. Geopolitical analysis, which for years has concerned itself mostly with explaining events, is increasingly concerned with describing how these events affect the physical world, in terms of supply chains, people, and resources. A war, an election, a regulation: these still matter, but they matter mostly for what they set in motion. Ships reroute, prices suddenly spike, workers cannot cross a border, components don't arrive.
Customers of geopolitical analysis in the 21st century rarely ask us "what happened?" Instead they ask "what does this mean for us?"
Why Traditional Intelligence Analysis Runs Behind the World
Answering that question means we have to change how analysis is done. Consequences do not wait for publication schedules. They arrive gradually and fragmentarily, all day, from every direction. An analytic process that engages with them only in concentrated bursts will always run behind the world it describes.
Consider how analysis has traditionally been conducted. In many intelligence shops, analysts arriving at a new desk "read in" on traffic and are then expected to generate analysis based on the summary of all that traffic in their head. Weeks of cables and reporting go in; a judgment eventually comes out. What happens in between is largely invisible, even to the analyst.
Analysts think of this process, the steady absorption of information and the gradual formation of a view, as "the real stuff" of the job. But in reality, the intake of information throughout the day and the expression of analysis that results from it form a black box that we don't understand very well. Why not make that process explicit, through tools that make it trivially easy for an analyst to engage constantly with the process of thinking about the world? Why not have analysis occur as a momentary, short response to each discrete event, in addition to periodic longer-form strategic assessments?
What Is Micro-Analysis?
This is why we place such value on what we call micro-analysis in a noisy, chaotic world. Micro-analysis refers to workflows that enable analysts to encode their thinking in response to events in small increments throughout the day, rather than in concentrated, irregular sessions when writing long-form products.
At Clock&Cloud, analysis begins with a basic building block we call the Clue. Clues are enriched signals: they capture events happening in the world that we care about and encode them with the semantic meaning and analytic judgment that an analyst possesses.
When a development crosses an analyst's desk, whether a ministerial statement, a strike on infrastructure, a new sanctions listing, or a shipping advisory, they log it as a Clue. Once logged, Clues contain a significant amount of proprietary data, which is what makes them valuable as pieces of micro-analysis.
Stitch enough pieces together, and you start to get a picture of the world. Stitch them together in inventive and interesting ways, and you can forecast what will happen in that world, and how it will affect physical goods and people moving around it. With enough Clues, and enough human creativity in the design and enrichment of the systems that use them, you can forecast scenarios, project losses, and even predict wars.
Why AI Makes This Possible Now
None of this is particularly revolutionary thinking. But in the era before AI, building a system to do this would have been cost-prohibitive, and maintaining it would have been a heavy burden on the analyst. The difference at Clock&Cloud is how much technology we can build and deploy to support the analyst through the massive expansion of rigor in their workflows.
Our systems handle the sheer volume of incoming reporting; the analyst supplies the judgment. Each Clue takes minutes. Over weeks and months, those minutes accumulate into judgments about the future.
None of this devalues the role of the human analyst. Instead, it makes explicit an implicit process that humans already engage in when processing information. Every analyst already weighs what they read, discounts one source and leans on another, and if needed adjusts their picture of the world. The Clue simply writes that adjustment down at the moment it happens, instead of leaving it to dissolve into memory. The analyst ends up with a diary of their thinking: an engram of their judgment, worldview, and perspectives.
Beyond the Finished Report
Working this way forces us to rethink what the output of analysis should look like.
For most of the discipline's history, the final output of analysis has been the finished product. An analyst could look back through papers they had written and point to the ones that held up well and those that aged poorly. But written papers and oral briefings lose the metadata of how judgments were derived. The alternatives considered, the sources doubted, and the signals that tipped the balance are all flattened into a few confident paragraphs.
Enhanced analysis captures not only the analytic bottom line, but also how we arrived at it, and which information and signals were useful in that process. With a digital trail of micro-analysis, we can interrogate all of that, and return to it whenever we need to evaluate it.
Beyond that, there is a second problem with products. In the digital age, products are rapidly decaying artifacts that lose relevance quickly, out of all proportion to how long they take to create. A single morning's news can overtake a report that took three weeks to write.
We need to think in terms of evergreen delivery of imperfect insight, constantly calibrated toward perfection. The rhythm of analysis for us is less that of the printing press than of the trading desk, where positions are marked to market continuously as new information arrives.
From Paper to Dashboard to MCP: How Intelligence Is Delivered
The shift is already visible in how intelligence is delivered. Analysis is moving from paper product, to notification, to dashboard, to MCP layer.
The paper product waited on a desk to be read. The notification pushed urgent developments to a phone. The dashboard offered a live view at any hour. The MCP layer closes the remaining distance. Built on the Model Context Protocol, an open standard that lets AI assistants connect directly to live data, it lets customers put questions to the AI tools they already use. The answers come back grounded in the latest analyst judgment, with the reasoning attached.
This is how information will be delivered between enterprises, and between people, in the future. Given the stakes of our work, it is important that geopolitical analysis is at the forefront of this trend.
At each stage, intelligence has moved closer to the moment of decision and further from the filing cabinet. The report on the desk is becoming a voice in the room: available at any time, with the latest context and the sharpest insight. That raises a question the old model was designed never to ask. If the analyst is now in the room, what role should they play there?
The Analyst in the Room: Why Analysis Has Always Been Advice
Every morning, in every analysis team, a set of subconscious decisions is made that nobody records. Which overnight development leads the brief, and which earns a single line? Whose statement is quoted and whose is ignored? Which risk is framed as imminent, and which goes unmentioned? By the time the brief reaches its reader, its most consequential choices have already been made, quietly, by the person who wrote it.
It is time to break a traditional government rule which says analysis should be divorced from policy-setting. This was a well-intentioned idea, rooted in the separation of military and civil relations in democracies and intended to keep politics in civilian hands and prevent deep-state coups.
The reality, though, is that analysis always involves advice. Selecting which information to value and present, which issues deserve focus, and how those issues should be covered is inherently a policy choice, and analysis teams make it every day. Maintaining that intelligence analysis is somehow apolitical or policy-agnostic is a fig leaf that poor analysis can hide behind when bad policy choices are being made that better judgment would strenuously advise against. When a decision goes badly, the analyst who "merely provided context" can always claim to have stayed in their lane, even when the warning that might have changed the outcome was muted, buried, or never written at all.
Because this choice is already being made implicitly, it is better to bring it into the light, where motives, bias, and frame of view can be made explicit for the end customer of analysis. Just as the Clue makes visible the many small judgments analysts previously formed in their heads, it also makes visible the biases analysts introduce into their products.
These biases are not pejorative. They are natural to the work of a human whose role is explanation, context, and advice. We should embrace these judgments and label them as such, rather than pretend they do not exist.
Reframing the analyst as an advisor, acting and advocating as an interpreter and guide for the customer, is desirable in an information environment of low trust and high noise. Amid that noise, decision-makers need a trusted guide willing to say plainly what they think, why, and how confident they are.
Trust Earned Through Performance: Scoring Geopolitical Forecasts
To achieve this, trust can no longer rest on reputation alone. It must be earned through performance.
Because we now preserve the record of micro-analysis, established, reputable, reliable advice can be measured and scored in a way that exposes charlatans and showcases strong analysts. Because judgments are recorded at the moment they are made, they can be checked against what actually happens, using methods such as Brier scoring, first developed to test weather forecasts. A Brier score measures how close probabilistic forecasts come to actual outcomes. An analyst who repeatedly calls outcomes 80 percent likely should be right about four times in five, and the record will show whether they are.
Critics will object that an analyst who advises will introduce unhealthy bias. The old rule never measured whether that happened; a scored record does, because a judgment bent to please eventually shows up as a miss, in plain view of the customer. The most rigorous analysts will rise, and their advice will reach decision-makers in the most useful phase of deciding, rather than as background context supposedly divorced from real-world outcomes.
AI in Intelligence Analysis: Only as Good as the Humans Deploying It
None of this is possible at scale without artificial intelligence. AI is here to stay as a tool and a partner.
It is still fashionable to point at mistakes made by AI and announce that they show the technology itself is failing. But like any technology, AI is only as useful and intelligent as the humans deploying it. A budgeting spreadsheet in careless hands produces confident nonsense too, but nobody concludes from this that arithmetic itself has failed.
Many AI hallucinations and errors signal poor alignment with the intended use of the humans working with them. Alignment errors are themselves a sign that the people using the AI are either unclear about their goals, or lack sufficiently deep expertise in their own fields to calibrate the AI to their purposes.
Here the black box returns. Even highly experienced analysts do not properly understand their work as a composite of many small, discrete tasks and activities. They think of it as a nebulous process by which many vague inputs result in a cohesive but messy picture of understanding.
You cannot hand a black box to a machine and expect it to replicate the process instantly. Aligning AI to analysis tasks requires a deep deconstruction of the analysis process, along with constant, rigorous mental work, both within oneself and in relation to the AI, to ensure its outputs reach the intended level of depth and quality. Our Clue workflow is just one product of that deconstruction: a single analytic habit, out of many, broken down and made repeatable.
At Clock&Cloud, this has settled into a consistent division of labor. The system handles scale, the analyst supplies judgment, and the system records and surfaces that judgment rather than manufacturing it.
An AI that generates poor results is an AI that has not been sufficiently calibrated to its role, given the right data, or instructed specifically enough. The specialist knowledge required to do these things, and to tell good output from bad, is part of the whole that will preserve human expertise in intelligence analysis.
In AI-enabled analysis, then, the durable advantage lies less in access to models than in the depth of domain expertise needed to deconstruct a discipline, align AI to it, and recognize when the output falls short.
What This Means for Globally Operating Companies
For companies with international supply chains, operations, and workforces, the consequences of geopolitical events arrive the way this article began: a slowed tanker, an overnight customs notice, an unexplained port delay. Intelligence that keeps pace with those consequences needs to arrive continuously rather than on a publication schedule, carry its reasoning with it, and be available through the AI tools teams already use.
It also needs to be accountable. When forecasts are recorded at the moment they are made and scored against what actually happens, decision-makers can judge the intelligence they rely on by its record, not by its reputation.
Conclusion: Where Weather Forecasting Stood a Century Ago
One hundred years ago, humans struggled to predict the weather with any reliability. Today, they expect a device in the palm of their hand to predict rain down to the minute. That leap came from meteorologists who understood their discipline well enough to write software describing it to a machine, and to correct the machine when it was wrong.
Geopolitical forecasting now stands where weather forecasting stood a century ago: looking up at a sky it cannot yet read, and holding, for the first time, instruments that might help it learn how.
Frequently Asked Questions
What is micro-analysis in geopolitical intelligence?
Micro-analysis is the practice of capturing analyst judgment in small increments throughout the day, in response to each discrete event, rather than only in periodic long-form reports. At Clock&Cloud, each unit of micro-analysis is recorded as a Clue.
What is a Clue at Clock&Cloud?
A Clue is an enriched signal. It records a real-world development, such as a ministerial statement, a strike on infrastructure, a sanctions listing, or a shipping advisory, together with the meaning and analytic judgment an analyst attaches to it. Combined, Clues support forecasts of scenarios, losses, and conflict.
How can geopolitical forecasts be measured for accuracy?
Because judgments are recorded at the moment they are made, they can be checked against actual outcomes using methods such as Brier scoring, which was first developed to test weather forecasts. An analyst who calls outcomes 80 percent likely should be right about four times in five.
What is an MCP layer for intelligence?
An MCP layer is built on the Model Context Protocol, an open standard that lets AI assistants connect directly to live data. It allows customers to ask questions through the AI tools they already use and receive answers grounded in the latest analyst judgment, with the reasoning attached.
Does AI replace geopolitical analysts?
No. In Clock&Cloud's model, the system handles the volume of incoming reporting while the analyst supplies judgment, and the system records and surfaces that judgment rather than manufacturing it. The domain expertise needed to calibrate AI and recognize good output from bad is what preserves the role of the human analyst.
How does geopolitical risk affect global companies?
Geopolitical events increasingly matter for their business consequences: rerouted ships, price spikes, workers unable to cross borders, and components that don't arrive. Companies with international supply chains and operations need intelligence that tracks these consequences continuously and explains proactively what they mean for the business.