ĢƵ

Accelerating Multi-INT Fusion for Intelligence Missions

Applying Technology for the Intelligence Mission

Knowing how to apply technology for the mission is key. “Everything we do drives toward more time, capability, or capacity for the government,” saysSaurin Shah, a leader in ĢƵ Allen’sartificial intelligenceand machine learning (AI/ML) business supporting the IC. “We help national security leaders find the right technologies and solutions around automation, digital modernization, advanced analytics, and AI/ML.”

“Everything we do drives toward more time, capability, or capacity for the government.”

Benefits of a Long-Term IC Partnership

Partnering with the IC over the decades allows us to see the tradecraft through practitioners’ eyes, giving insights to help clients modernize in tangible ways. Open-source intelligence—bringing together publicly available data such as social media—is one example.

“We worked side by side with government clients to help them develop the legal policy for working with social media,” says Rob. “We need to protect U.S. citizens’ privacy while finding those who threaten their safety.”

A critical issue we helped clients codify into policy is collaboration security, being able to trust the data. “Veracity of the source is important. You can grab all kinds of things and probably tell any story that you want to tell, but if you're using sources that are unreliable or deliberately misleading, you’ve got a problem,” says Rob.

“That extends to the supply chain for the data source and the history of the data. You need to be able to show the breadcrumbs of how we got to that rationale. That’s part of our legacy, helping clients build and refine their policies over time.”

Fusing Macro and Micro Approaches

Advancing multi-INT fusion requires a multiprong approach. The intelligence community needs quick wins that work with legacy systems while moving toward sweeping transformation.

We supportlarge-scale modernization; for example, helping move defense organizations toopen frameworks. Our teams also provide cloud solutions for intelligence clients—the classified equivalent of our work in helping theU.S. Treasury stand up a cloud environment with the highest security level.

At the same time, we innovate ways to give rapid benefits that work with legacy systems today and will integrate seamlessly with the open frameworks of tomorrow.

Legacy Systems: Quick Wins

“Automation tools can be as simple as recommenders,” says Rob. “When you go to Google and start typing words, it’s going to recommend things to you. We’re bringing that to a classified environment.”

Simple as a recommender appears to the user, it’s powered by sophisticated algorithms, seamlessly making connections between multiple sources. Based on the user’s search patterns, this combines activity-based analysis, which focuses on finding relationships based on behavior, with object-based analysis, where attributes of images or descriptions are analyzed to reveal correlations with what’s being searched.

“This helps algorithms learn what the analyst is looking for,” says Rob. “The algorithm sees an item in a SIGINT [signals intelligence] report, which is potentially the same item that’s being referenced in this imagery report, and brings it together in an automated way for the analyst.”

Micro applications like this save analysts time and can be rapidly integrated into legacy systems. “Quick scripts, quick algorithms can solve an immediate problem and may be relevant to something else. It doesn’t cost a lot, doesn’t take up a lot of bandwidth. The government can ‘toss the code’ and go on to something else,” he says.

Case Study: Documentation Innovation

Intelligence analysts must document their assessments in detail for reasons such as continuity, transparency, and auditability. It’s important for being able to trace the path behind a recommendation and becomes critical in cases where a source is later found to be unreliable. Creating these records is tedious and takes time the analyst could be using to solve problems close to the mission.

For one IC partner, a ĢƵ Allen team developed an automated source extractor capability that automatically appends an analysts’ data with the source—reducing the time it takes an analyst to properly cite sources by 75%.

Modular Transformation: Accelerating Results

“We’ve had a lot of success supporting national security missions through the integration of AI/ML and advanced data science techniques such as probabilistic and predictive modeling and automation of data pipelines,” says Saurin. Our work increases the pace of analytics to machine speed while streamlining analytic processes. He likens the experience to SIRI. “You still take notes, right? But making that to-do list is a lot easier.”

AI/ML enables collection and exploitation, drawing connections at lightning speed that advanced AI can analyze to predict outcomes and recommend action. Target activities can then be correlated to create fast decisions where every second counts.

Another major development area, data engineering and automating data pipelines, saves analysts time and resources. “Our work facilitates not only the ingestion, but also the conditioning and enrichment of that data for analytic value. That gives data scientists and analysts so much time back,” he says.

Escalating Multi-INT Advantages

As intelligence organizations modernize, we create more sophisticated cross-domain pipelines. “A lot of new sensors are coming out there. Fusing those new classified streams with commercially available and open-source data at scale requires new, layered levels of data engineering and analysis,” he says.

The data needs to be indexed, fused, and analyzed at appropriate security levels for multiple stakeholders, from the intelligence community to military, government, and international partners. Then those recommendations need to be distilled to send to leaders at their level of access, at the moment needed.

Such large-scale, diverse fusion requiresopen architecturesthat move operations to the cloud for bandwidth and scale. To work within allotted budgets, many intelligence clients need to approach thatlarge-scale modernizationin terms of iterative changes that build toward a larger vision.

Open-Source Solutions Simplify Modernization

We solve the clients’ dilemma with a modular open-architecture approach, which simplifies transformation while allowing clients to proceed according to their funding resources. “We’re agnostic to the solution, so we’ll choose the best technology for that organization,” he says.

Often the answer lies in tools and frameworks we’ve already created from commercial off-the-shelf or open-source software—tools that are already cleared through the accreditation process, have intelligence-level cybersecurity built in, and are proven in operation.

“The right technologies and solutions accelerate decision making while simultaneously providing capacity,” he says. “The beauty is that it all works together.”

Case Study: Platform One Enables First AI Co-Pilot

To keep the U.S. ahead of escalating threats, ĢƵ Allen partnered with the Air Force to developPlatform One, the federal government’s first enterprise-level DevSecOps service. The service standardizes acquisition, policies, and platforms so developers can focus on creating the custom elements. Using an open, modular infrastructure ensures flexibility while automating repeatable processes speeds delivery. Using Platform One, the partners operationalized an AI algorithm within weeks—successfullydemonstrating human-machine teaming aboard a Lockheed U-2 aircraft, a first for airborne reconnaissance missions.

Get Insights

Fill out the form below to get the latest insights on national security from us.