Speed Read
- To address urgent national security challenges around the world, intelligence community (IC) analysts and warfighters need to make sense of complex, uncertain situations—and to do that, they need next-generation sensemaking systems that can collect and analyze data on an exponentially increasing scale.
- These tools can quickly synthesize data, present options to analysts, and operate as assistants. Imagine, for instance, new algorithms designed to strengthen collection, more powerful sensors with greater analytic power at the edge, computer vision models for capturing new objects or performing change detection in key areas, large language model (LLM) agents to ease the challenge of sifting through massive amounts of data, and analytics services that fuse commercial signals and satellite imagery.
- It’s time to take action. Organizations can create next-generation sensemaking systems now by assessing the current state and building a roadmap, undertaking pilot modernization projects, and accelerating modernization at scale.
Next-Generation Systems Can Meet Urgent Needs Now
At dawn, Grace, an all-source intelligence analyst for a three-letter agency, logs in to her workstation at headquarters. From her screen-filled desk inside the Beltway, she turns her top-secret gaze to the Indo-Pacific region, half a world away. It’s the same routine she’s had for three years in the role. But this morning, February 3, 2026, is different. She sees an urgent email from her supervisor. What comes next, she can’t stop. All at once, her brows arch up, her eyes widen, her jaw drops, and she gasps.
But the headline is clear: “China Buys Uninhabited Fijian Island to Build Military Base.” The news comes from the Beijing bureau of a major U.S. daily. “A Chinese development firm with links to the Chinese government and People’s Liberation Army today announced that it recently purchased the uninhabited Cobia Island from the government of Fiji for $850 million,” the story says. “Western security analysts assess that China plans to use the island to build a permanent military base in the South Pacific, 3,150 miles southwest of Hawaii.”
The news means the People’s Republic of China (PRC) can boost its power projection in the Pacific Islands, a region of strategic importance, expanding its military presence near U.S. and Australian territory.
It’s the kind of surprise that Grace and her intelligence community (IC) colleagues always try to avoid. They know in their gut they must focus on hard targets to collect information on adversary plans that would otherwise go undetected, just as a Center for Strategic and International Studies (CSIS) report recommended. A hard target is “a person, nation, group, or technical system” that poses a potential threat and a very difficult intelligence-gathering challenge—it’s “often hostile to the U.S. or heavily protected, with a well-honed counterintelligence capability.” In this case, the PRC succeeded in keeping the U.S. in the dark.
“It’s why we call it the pacing challenge,” Grace tells herself half-heartedly. It’s no consolation.
The Truth in the Fiction
This imaginary scenario starts with a fictional headline borrowed from a 2017 National Intelligence Council global trends report. Behind our make-believe scene is the IC’s real need to get ahead of surprises by accelerating its ability to collect, analyze, and make sense of vast amounts of data. This challenge is intensifying as the number and quality of sensors rises around the world.
Although Grace’s gaze is set on the Indo-Pacific, no foreign power has a monopoly on inscrutability. Decades ago, Winston Churchill described the challenge of forecasting Russian actions as “a riddle wrapped in mystery inside an enigma”—and that still sounds about right. To address the full range of national security challenges around the world, IC analysts and warfighters need to make sense of complex, uncertain situations—and to do that, they need sensemaking systems that can collect and analyze data on an exponentially increasing scale.
Grace is a stand-in for real all-source analysts tasked with monitoring PRC foreign influence. All-source analysts review and analyze all kinds of data to assess, interpret, forecast, and explain a range of national security issues and developments that are regional or functional in nature. They use the full complement of all-source tools, techniques, and procedures to fuse all available intelligence into a comprehensive product designed to satisfy unit-level essential elements of information (EEI) and command-level priority intelligence requirements (PIR).
Sensemaking is the fundamental cognitive ability human analysts use to understand emerging developments and anticipate future events. It’s not the work of a lone analyst but happens with the fusion of many viewpoints. Increasingly, technology plays a supporting role. A decade after the 9/11 attacks, the Intelligence Advanced Research Projects Activity (IARPA) described sensemaking as “the remarkable human ability to detect patterns in data, and to infer the underlying causes of those patterns—even when the data are sparse, noisy, and uncertain.” IARPA also called for the development of new automated analysis tools designed to replicate the strengths of this ability. A 2019 IC strategy for augmenting intelligence using machines defined sensemaking as “a process of creating understanding in situations of high complexity,” urging the construction of shared models “to provide the basis for trust between human and machine teams.”
Had the team in our hypothetical scenario been able to comb through vast amounts of data more efficiently and effectively, they might have uncovered the brewing PRC-Fiji deal earlier, enabling the U.S. to proactively manage related risks and perhaps even shape the outcome.
In this first version of the story, the sensemaking capabilities at the team’s disposal aren’t as robust as they need to be. The IC is still moving from counterterrorism-focused approaches to the new technical capabilities designed for collection and analysis against hard targets and the pacing challenge. Now let’s reimagine our scenario, but this time with transformed sensemaking systems already in place.
Sensemaking Reimagined
Rewind six months. Grace is sitting across from her supervisor in a nondescript meeting room at headquarters. An initial review of collected signals carried out by a fine-tuned large language model (LLM), a type of artificial intelligence (AI), has triggered a high alert involving hard targets. Now human analysts need to investigate.
“The PLA and this development firm are talking. No idea why,” the supervisor says, pushing a file to Grace. “We need to identify any potential operations. That’s where you come in.”
Grace looks at the photos. The development firm is led by a notorious “retired” PRC intelligence officer, a very slippery target indeed—rarely seen, never overheard. The other few targets are People’s Liberation Army (PLA) leaders who work in highly secure facilities.
“You’ll also have access to these resources,” the supervisor adds, listing tools with sensitive unrepeatable names. The conversation doesn’t dwell on how the tools include new algorithms to strengthen collection, more powerful sensors with greater analytic power at the edge, computer vision models for capturing new objects or performing change detection in key areas, LLM agents to ease the challenge of sifting through massive amounts of data, and analytics services that fuse commercial signals and satellite imagery. At this moment, all that matters is the tools will help Grace and her team quickly collect the right data and connect the dots. The tools will quickly synthesize the data, present options to analysts, and operate as assistants.
For the intelligence collection, Grace targets key individuals, their organizations, and relevant technical systems (e.g., devices and equipment used by the targets). Weeks go by. They meet again in a hurry when Grace reports a breakthrough. Several targets are talking to each other about Fiji, with one even naming a specific location: Cobia Island. Shaped like a crescent moon and lined with tall trees, the island is a submerged volcanic crater. Intrepid tourists visit for hiking, snorkeling, and kayaking.
“Maybe the PLA would rather park warships there,” the supervisor says. “We’ll need new overhead satellite imagery.” The team assembles more collection plans and starts putting them into action. As it turns out, the AI assistant has already anticipated the need, and has ordered and received commercial satellite imagery of the island. But there is a brief setback, a glitch.
In the overhead imagery, the surrounding waters are clear, but part of the island keeps coming back fuzzy, almost like it’s camouflaged. Maybe it isn’t a glitch. Computer vision analyzes the pixels and Grace adjudicates. The target becomes just clear enough to confirm someone is skillfully hiding something. But what?
Grace turns to a teammate, Connor, a collection manager who helps all-source analysts get the intelligence data they need to answer key intelligence questions. Seeing the need to collect more imagery as well as electronic signatures in the area, Connor tasks a drone to get a better look from multiple angles.
The new 3D imagery arrives. Grace opens the file. All at once, her brows arch up, her eyes widen, her jaw drops, and she gasps.
“Aha!” A new structure is hidden in the breathtaking beach forest. The building has an uncommon design. The footprint is relatively small but perhaps the start of something bigger. What’s more, there are signs of a recent oil or chemical spill at the site. It’s unclear if the people there have noticed the spill yet. Even if they have, they may not know the contamination is spreading behind the structure down to the white sands and crystal waters.
Delving deeper, the team uses machine learning (ML) algorithms to parse signals from the area to see if they can be tied to known actors. The results show personnel from the shadowy development company are likely at the site. These people are overheard saying the PRC plans to buy the island. Grace’s supervisor is elated.
“If I was a citizen of Fiji, I’d want to know about pollution threatening the pristine environment,” the supervisor says. “Great work.” Somehow, leaders in Fiji’s tourism industry and environmental community soon learn there’s been a spill on the island. Before the PRC-backed development firm can close a deal to buy the island, the public in Fiji voices strong concern, changing the calculus for Fiji’s government, which no longer feels comfortable making a sale. What’s more, the team’s work has established a pattern of life for the collection targets. In the future, the IC can use this pattern to identify and get ahead of similar risks elsewhere in the region.
The need for next-generation sensemaking isn't over the horizon-it's urgent
Transforming Sensemaking Capabilities
This is just one hypothetical scenario involving hard targets. Others might look very different. The need to identify, analyze, and track emerging developments around the world is urgent. The U.S. military’s Joint All-Domain Command and Control (JADC2) strategy prioritizes sensemaking to turn data into information and information into knowledge through “the ability to fuse, analyze, and render validated data and information from all domains and the electromagnetic spectrum.” And in the unfortunate event of a military crisis or conflict, sensemaking capabilities would rapidly scale in certain ways. But there is no time to wait. More and more, threats from nation-states are taking the shape of coercive or subversive actions below the threshold of armed conflict. Scaling and strengthening sensemaking capabilities now is crucial for national security and global stability.
Overall, the IC needs to adopt, buy, or build the right technology to transform their sensemaking capabilities so they can collect against and analyze hard targets despite advanced adversarial surveillance and counterintelligence capabilities. The nature of near-peer conflict makes this continued modernization all the more difficult. These systems and advanced analytics need to work in disrupted, disconnected, and intermittent low-bandwidth (DDIL) environments. These constraints require new approaches for development and operations. For instance, a CSIS report recommended using forward collection teams to push AI-enabled collection and analysis closer to operators in contested areas. Pushing these capabilities to the edge is essential for maintaining the advanced analytics required, even as adversaries work to disable U.S. mission systems.
In addition, the IC needs scalable systems with more computing power than ever before to churn through data and prepare insights for policymakers and warfighters. Otherwise, because current manual systems only let teams analyze so much data at a time, intelligence is left on the table—and that intelligence could potentially be decisive in a crisis or conflict. Imagine, then, the national security value of having algorithms parsing through vast amounts of data and accurately alerting scientists to information of interest for further investigation.
But this isn’t just about technology. What’s needed is comprehensive modernization across the enterprise, including strategy, governance, culture, talent, data, technology, and IT processes. To that end, the Defense Innovation Board has recommended “aligning incentives to drive faster tech adoption” by embracing risk, providing “top cover,” stopping the tendency to reward mediocrity, accelerating innovation and tech development, creating a career path for innovators, tracking people innovation readiness levels, defining a vision for innovation, and creating a culture of learning and innovation.
By pursuing mission-driven innovation, your organization can focus not just on quickly buying the latest tech but on accelerating mission outcomes. You can more rapidly adopt key technologies that address current and expected requirements, maximize the likelihood that investments will lead to the fielding of suitable and effective capabilities at mission speed, factor in tradecraft and workflow considerations from the start, proactively consider the ability to scale over time and meet future processing needs, and mitigate the risk that fielded capabilities will quickly become obsolete.
The Defense Innovation Board called for a better understanding of how industry works, greater business acumen, and tighter collaboration with providers of cutting-edge tech. Industry partners who bring a mix of deep mission understanding and key connections with other leading providers of crucial tech—for example, cloud computing, AI, and ML—can enable the IC to create more resilient and responsive sensemaking at scale. Imagine, for instance, onboarding hundreds of new models to augment and improve tradecraft, delivering a clear advantage to policymakers as well as personnel in the field.
The need for next-generation sensemaking isn’t over the horizon—it’s urgent. CIA Director Bill Burns has said how well or how poorly the agency takes advantage of emerging tech and transforms its tradecraft—to stay ahead of adversaries—will “make or break us as a professional intelligence service.” The same holds for the IC as a whole. It’s time to take action. To start creating next-generation sensemaking systems today, your organization can use the three steps below. Together, the IC and industry can address this challenge.
3 Steps to Create Next-Generation Sensemaking Systems Now
1. Assess
- Assess your current state and build a roadmap. Use a holistic framework that empowers organizations to understand, grow, and reinforce their analytics capability. Identify strengths, gaps, and needs across multiple areas, including strategy, governance, culture, talent, data, technology, and IT processes. Build scorecards to prioritize areas for modernization. Develop and release a strategic vision for modernization. Include clearly defined, achievable goals and success metrics. Assess the maturity of current zero trust cybersecurity capabilities. Build a modernization roadmap with necessary actions, timelines, and resources to meet short-term and long-term modernization goals with measurable results.
- Focus on data. Data is key for sensemaking. Modernize back-end data storage technology for a legacy system to improve data management, integration across stovepipes, accessibility, and availability, and to enable ML workflows.
- Enable ML operations. Expand on modern data architecture and data-mesh concepts to set up ML pipelines for predictive modeling with continuous evaluation. Evaluate best-of-breed foundation models to synthesize large amounts of data. Establish processes to enable multiple models to pick the right tool for the right job. Don’t put all your eggs in one basket. Be ready to constantly update. An example of a success metric would be “achieve 98% data availability.”
- Leverage leading DevOps tools. Use a deployment framework that includes open-source clusters for hosting containerized software. Establish AI-enabled development environments with low-to-high continuous deployment pipelines to increase velocity and agility. Include program health monitoring and metrics.
2. Activate
- Conduct pilot modernization efforts. Include change-management activities. Maximizing the understanding, willingness, and capability of the workforce to move from the current state to the future state is crucial for risk management. Identify a system or process to execute modernization on. Start with something that is limited in scale and noncritical.
- Start modernization. Conduct modernization activities on the selected system/process. Align success criteria to real mission impacts, focusing not on the technology but on what that technology enables. Focus on iterative delivery and integration into the mission. Highlight mission outcomes and measurable impacts for key stakeholders. Provide training and engage users. Document new tradecraft and workflows. Informed by the earlier assessment, adopt zero trust capabilities to enable data-centric security, data sharing across security enclaves, and greater interoperability. Share the results with key stakeholders. If, for instance, you succeed in modernizing your infrastructure to do advanced analytics at scale, leadership will likely want to know.
- Introduce digital assistants. Introducing AI-enabled digital assistants to supercharge analyst productivity is an example of potential modernization activities. Build them into common working environments and workflows to proactively pull in applicable research and provide preliminary analysis so humans can focus on higher-value activities.
3. Accelerate
- Scale modernization efforts. Modernize critical systems and infrastructures iteratively. Measure and monitor performance and optimize. Develop documentation and enable knowledge sharing. Standardize tradecraft and workflows for accomplishing the mission across the organization. Continue user engagement, feedback, and integration activities. Across all phases, maintain clear and consistent communication and manage expectations. Be prepared to adapt the roadmap based on new insights, and manage and alleviate risks.
- Build AI agent functionality. This is an example of scaling modernization. Prototype and begin phasing-in AI agent functionality that can autonomously—with supervision—complete routine intelligence workflows by interacting with knowledge bases, tools, and external systems to meaningfully increase productivity, efficiency, and time to insight
VELOCITY MAGAZINE
Booz Allen's annual publication dissecting issues at the center of mission and innovation.
