The Future of Corporate Intelligence Part 3: Predictive Intelligence

Predictive intelligence - forecasting what could be coming up, rather than reacting to what has happened - can unlock significant gains for intelligence teams, but only if it drives action and is plugged into the right processes

Introduction

Many intelligence teams get stuck firefighting, constantly responding to things that have already happened. They spend valuable time writing reports on the impacts of past events, even though their real value lies in forward-looking assessments and forecasts that help the business stay ahead of emerging threats. Shifting focus, however, requires time, space, expertise, and a significant cultural change.

Yet, it is clearly in a business's interest to prioritise predictive intelligence. It offers a clear route for intelligence teams to add business value rather than be seen as a cost centre. Those teams that successfully leverage forecasting models and predictive intelligence will demonstrate the greatest return on investment.

The good news is that it's easier than ever to research and build forecasting models. The computing power of AI, combined with vast and growing datasets, and a suite of successful models already tested in academic and governmental contexts, has democratised access to predictive capabilities once reserved for governments or elite institutions.

Let’s get some things straight at the start. Predictive intelligence can sound far fetched if not caveated appropriately. Using AI to predict the future sounds like something pulled from the recycling bin of the Black Mirror writing room. But when we’re talking about predictive analytics in intelligence, we’re just talking about models that help us better understand trends, identify early-warning signals, and make smarter decisions in a world that feels increasingly unpredictable. It’s work analysts are already doing - just amplified by technology. As computing power and data availability grow, so does the potential.

Real-World Applications and Success Stories

Despite how chaotic politics and society can seem, people don’t act randomly. Certain conditions consistently increase the likelihood of instability, and there are patterns to human behaviour.

Many of the drivers of violence and unrest are intuitive and evidence-backed: income inequality is linked to violent crime, and youth unemployment correlates with participation in political violence. Environmental factors also play a role. A 2023 study found that higher temperatures correlated with increased violence and self-reported intent to commit violence in Iraq and Afghanistan.

Governments, political scientists, and data scientists have all built models to predict future outcomes - some with measurable success. The Raven Sentry project, run by the US government in Afghanistan from 2019 to 2020, used open-source data to forecast insurgent attacks. The model reached 70% accuracy in predicting when and where attacks were likely to occur, a level akin to human intelligence analysts, but achieved in a fraction of the time.

Better statistical modelling has been able to more accurately anticipate food crises outbreaks (a known driver of instability and conflict), even if previously these events were thought to be difficult to model and predict.

Political scientist Bruce Bueno de Mesquita developed an Expected Utility Model (EUM) that forecasts policy decisions by simulating bargaining scenarios among political actors. The model has been credited with accurately predicting various political outcomes, including leadership successions and treaty negotiations. For instance, it correctly forecasted the succession of Indian Prime Minister Y. B. Chavan and the subsequent political dynamics. He says of his approach “Certainly knowing about places and how different they might be is important, but not as important as knowing about people and how similar they are, wherever they are.”

While much of this work remains in academic or governmental domains, there are clear implications for corporate intelligence teams. As AI opens up access to sophisticated models, the most successful corporate teams will be those able to harness forward-looking insights to inform action.

Limitations, Challenges, and Ethical Landmines

The potential of AI powered predictive intelligence is compelling, but it carries significant limitations that have to be handled carefully.

Complacency and over-reliance on the model: Analysts must remain central to the process. Predictive models should augment, not replace, critical thinking. Any model should be used to augment and expand their work, but not become a crutch. Even experts in their field can be influenced by the results of AI analysis, as seen in a study which found even experienced radiologists had their diagnoses swayed by intentionally incorrect AI recommendations.

Bias and Ethical Concerns: Predictive models can inadvertently perpetuate biases present in historical data. If past data reflected biased enforcement or reporting (as is often the case in law enforcement or political reporting), then predictive systems may reinforce those patterns. These concerns lead Santa Cruz to be the first US city to ban predictive policing. Explaining the reasons behind the move, the mayor of the city, Justin Cummings, explained "We have technology that could target people of colour in our community - it's technology that we don't need.”

Data challenges: Any predictive model is only as good as the data that supplies it, and sometimes the data needed to make a meaningful assessment on likelihood is local, and difficult to collect unless you have an on the ground network. As Jonathan Bellish, executive director at One Earth Future’s told the Washington Post, “In one country…we found that the possibility of violence could be correlated to the number of dogs outside, because worried people would pull their dogs in off the streets. That’s a very useful data point. But it’s hyperlocal and requires knowing humans on the ground. You can’t build that into a model.”

Also, a lot of models aren’t that good, and there is the danger of decision makers being persuaded by a charming snake oil salesman with a fancy dashboard.

The Analyst’s Role in an AI-Augmented World

The strength of predictive analytics lies in how it enhances an analyst's capabilities - not in replacing them. Analysts who can build effective models, monitor key triggers, and integrate contextual cues will produce the most meaningful insights.

With AI handling data at scale, human analysts can focus on asking better questions, layering nuance, assessing implications, building relationships and delivering clear advice to stakeholders. Predictive intelligence, when embedded properly, turns analysts into strategic advisors who help their organisations act early, not late.

What Can Corporate Teams Actually Do With It?

A well-designed predictive model acts like a second set of eyes - tirelessly scanning for emerging risks. But models only add value if they lead to action. Intelligence teams still play a critical role in ensuring those insights reach the right people in the right way.

This is where predictive intelligence often stumbles. Many companies like the idea, but struggle to operationalise it. That’s why models and forecasts must be tightly integrated into existing workflows and geared toward answering specific questions or monitoring clearly defined scenarios.

Producing a three-year outlook might look impressive, and serve as a good internal talking point, but it rarely prompts meaningful action. Effective predictive intelligence should inform real decisions, not just decorate PowerPoint slides.

This also raises a key question: what is the right forecasting horizon? If you look too far ahead, insights can become vague and impractical. Focus only on the next few hours, and you miss the window to act strategically. The right time horizon depends on the nature of the risk and the internal actions it supports.

For example, assessing future developments in the Russia-Ukraine war might suit a six-month outlook, given the complexity and slower-moving nature of the conflict. In contrast, evaluating the risk of full-scale war between India and Pakistan during a period of high tension might require a much shorter horizon, perhaps just a few weeks, as the threat landscape can change rapidly.

Conclusion

The future of corporate intelligence won’t be defined by machines replacing analysts. It will be defined by analysts who know how to use machines wisely.

Predictive intelligence isn’t about knowing the future. It’s about being less surprised by it.

When it’s used well, it sharpens foresight, drives action, and gives companies an edge in an unpredictable world.