Presented by Carahsoft
From zero trust and artificial intelligence assurance to data-centric modernization and cyber intelligence, leaders from across the intelligence and defense communities are rethinking how technology supports the mission. Recorded at DoDIIS 2026, these conversations explore how government and industry are moving beyond collaboration toward deeper integration—while keeping security, trust, workforce readiness and operational outcomes at the center.
Shrader says global competition is accelerating that shift. U.S. adversaries are moving quickly not only because of advances in technology and computing power, but also because of the way their governments and industries work together. That reality is pushing federal leaders to acknowledge more openly that commercial companies may be ahead in important areas and that government needs industry’s capabilities to solve its hardest problems.
Effective integration begins with how products are designed. Open architectures allow agencies to add or replace modular capabilities without rebuilding an entire environment. Secure-by-design products make it easier to introduce technology into sensitive networks because security is embedded at the architectural level. Those technical attributes must be matched by a culture of trust: government must be willing to learn from industry, and industry must understand and support government missions.
Artificial intelligence makes modularity particularly important. No single company is likely to supply every component of a complex AI architecture. Open systems allow agencies to combine specialized technologies, insert new capabilities as they emerge and automate parts of the intelligence process. Shrader says early results are validating that approach by accelerating analysis, data fusion and the broader intelligence cycle. He adds that these lessons extend beyond the intelligence community to civilian agencies, state and local governments, education and commercial organizations.
Key Takeaways
That identity-centered approach must now extend beyond people. As NGA introduces agentic AI, autonomous software agents will interact with systems and information in ways that resemble human users. Chatelain says those agents require credentials, defined permissions and access decisions of their own. Before an AI capability enters the enterprise, NGA wants to understand its purpose, intended behavior and access to personal or mission data. Responsible-use policies established by the chief AI officer provide a governance framework for those decisions.
NGA is progressing toward intelligence community zero trust targets, including basic and then advanced maturity. Because the agency has identified its systems and data, the next step is applying the appropriate controls throughout the environment. Chatelain notes that the threat continues to evolve, including the possibility that quantum computing could undermine current encryption. Preparing for post-quantum security is therefore part of maintaining a healthy long-term defensive posture.
People remain central to the strategy. Chatelain identifies workforce training as one of his most significant concerns because employees need to understand AI, cybersecurity and the risks of exposing sensitive information. NGA has established prompt-writing training to help employees use AI effectively and responsibly. The larger lesson is that secure adoption requires technology controls, governance and a workforce capable of making sound judgments about rapidly changing tools.
Key Takeaways
Matthew describes four priorities. The first is resilient connectivity: without it, even advanced devices and applications become unusable. The second is content security, ensuring that intelligence reaches the correct warfighter with the right entitlements. The third is accelerating capability delivery. Commercial users can install a new application in minutes, while government delivery can take years—an unsustainable gap when AI models and other technologies change within months. The fourth is content discovery, supported by a data environment in which information is tagged, cataloged, encrypted and governed by entitlements.
The modernization effort must also account for AI-driven cyber risk. Matthew warns that autonomous agents could identify and exploit vulnerabilities at a speed and scale beyond human attackers. Agencies therefore need governance and identity controls that give agents access to what they require without allowing them to move beyond their assigned purpose.
Together, these priorities reflect a mission-centered view of IT. Modernization is not simply a refresh of infrastructure. It is a way to keep DIA’s intelligence available, discoverable and secure under pressure while closing the gap between commercial innovation cycles and government deployment. The outcome is a more resilient digital environment capable of supporting warfighters and preserving military advantage as both technology and threats accelerate.
Key Takeaways
Once data is visible, classification becomes the backbone of secure collaboration. Organizations must understand the sensitivity of information and apply the markings that determine how it may be shared. Grancarich says classification remains an area of substantial demand because it influences every downstream step—from protection and transfer to curation and access.
The goal is not simply to collect and label more information. Agencies need to identify which elements matter for a particular decision and bring structured and unstructured sources together into a coherent view. That holistic approach allows an operational team to use the most relevant data at the moment it is needed instead of navigating disconnected systems or relying on point solutions that do not work together.
Grancarich describes an interoperable model that connects discovery, classification, protection and data transfer. Together, those capabilities help organizations maintain control while enabling information to move securely. Curation then reduces the volume to the data most useful for the mission. The result is a foundation for secure collaboration and greater decision speed: leaders gain a fuller picture, analysts spend less time locating and reconciling information, and teams can act with confidence based on data whose sensitivity and relevance are understood.
Key Takeaways
Baker compares that model to how an agency develops a junior analyst. Leaders do not send a new employee’s first draft directly to the Secretary of Defense. They trust the doctrine, procedures and layers of review that shape the analyst’s work. AI needs an equivalent assurance environment—one that makes risks visible and gives leaders confidence to authorize responsible use.
The workforce remains decisive. As AI reduces the time and effort required for research, writing and other forms of execution, the most valuable skills become subject-matter expertise, critical thinking, asking the right questions and judging whether an answer is useful. DIA is therefore focused on AI fluency rather than narrow proficiency with one tool. Employees should understand what AI can do, where it can fail and how much of a workflow they are comfortable assigning to it.
Adoption cannot be sustained through mandates alone. Baker favors making approved tools easy to use, sharing examples from employees who have achieved meaningful results and allowing comfort levels to develop. Some users will adopt quickly and may need more skepticism; others will wait until the technology is well established. A responsible strategy accommodates that curve while building trust through real mission value.
Key Takeaways
Trust remains essential as systems become more advanced. Actionable intelligence cannot come from an unexplained black box. Analysts and decision-makers need to trace findings back to the underlying information, examine potential connections and corroborate results. That transparency strengthens confidence without eliminating the human review that high-consequence decisions require.
The urgency is heightened by adversaries using similar technologies without the same bureaucratic barriers between adoption and execution. Wallach argues that government must streamline the path from emerging capability to operational use so it can move at the speed of the mission. Removing unnecessary friction does not mean removing safeguards; it means making accreditation, integration and review responsive to rapidly changing technology.
The real promise is not simply doing the same work faster. AI can reveal connections analysts could not see before and compress work that once required a week into minutes. That time savings can create room for reflection and validation before intelligence reaches a commander or operator. Success is measured by whether the information arrives early enough to change an outcome and protect lives—not by the speed of the algorithm alone.
Key Takeaways
P
That requires business process reengineering as well as technical modernization. Many systems were built two decades ago and reflect processes that may no longer fit current operations. Grimsley says the command is examining the data feeds and processes across those systems so it can turn raw data into information leaders can use to set the globe and the theater.
The department already has abundant data, much of it unstructured. Government should pair mission experts who understand that information with leading commercial, academic and laboratory talent. AI can then help encode expert knowledge, automate repetitive work and become a force multiplier. Grimsley tells his team to “embrace your inner laziness”—using AI to reduce unnecessary effort while still applying thought and understanding.
Success includes quantitative measures such as technology cost savings and qualitative operational outcomes such as reducing planning from eight or nine hours to three or four. Cybersecurity must be built in from the beginning, particularly as agentic AI and AI-assisted software development introduce new risk. USTRANSCOM’s end state is faster, more resilient decision support that allows people and technology to concentrate on moving forces and resources where they are needed.
Key Takeaways
Scheibe emphasizes outcomes over products. AI is a tool for achieving a mission result, not a solution by itself. Its value depends on trusted existing data. Organizations should know where their data resides, maintain consistent access, identify overlap and improve fidelity before expecting AI to produce deterministic, useful decisions.
Modernization does not necessarily require discarding legacy systems or their data. The more immediate problem is that valuable information often remains divided across organizational and technical silos. Agencies can modernize how they access that data and layer AI-enabled capabilities above existing sources to connect the silos and accelerate decisions.
Governance remains essential. Faster access does not eliminate ownership, permissions or controls. Agencies can retain user-based access rules even when AI agents help employees work across data sources. The objective is to connect information without surrendering control, giving authorized decision-makers timely access to the data they need.
Key Takeaways
The work requires continuous baselining because cyber conditions are highly temporal. Firmware, addresses and configurations can change quickly, while crises frequently emerge in unexpected locations. DIA’s global overwatch mission therefore depends on building the best available picture before a crisis so analysts are not starting cold when events escalate.
The larger challenge is exploiting an overwhelming volume of cyber data. Analysts once asked for more information; today the volume of malicious activity is beyond what any human team can process. Allen sees AI, machine learning and automation as essential for reducing the noise and prioritizing what analysts need to examine. Discovery is useful, but curation—determining what matters now, later or not at all—offers the greatest value.
Automation cannot replace analytic rigor. Models must be understandable rather than black boxes, and quality assurance must give leaders confidence in how an assessment was produced. Analysts with deep problem-set knowledge must ask the right questions, evaluate outputs and remain in the loop for consequential decisions such as targeting. The objective is a human-machine team that combines machine scale with human context, rigor and accountability.
Key Takeaways