The hard problem isn't building smarter agents—it's making them safe
In the long arc of human endeavor, the institutions that cannot afford failure have always demanded the most rigorous architects of trust. Mark Ryland, a veteran of three decades building secure infrastructure at Microsoft and Amazon Web Services, has joined Arlington-based Istari Digital as Vice President of Artificial Intelligence — bringing his career's defining question with him: how do autonomous systems earn the right to operate where the cost of error is measured not in revenue, but in consequence. The appointment reflects a broader reckoning in defense and aerospace, where the promise of AI agents collides with the immovable reality that sensitive data cannot simply be gathered in one place to make machines smarter.
- Defense and aerospace industries are racing toward AI agents, but face a constraint consumer tech ignores entirely: classified and proprietary data must stay compartmentalized even as AI systems need to work across those very boundaries.
- The tension is not about intelligence — the market has no shortage of capable AI — but about whether autonomous systems can operate safely, accountably, and without demanding the kind of data consolidation that would breach security protocols.
- Istari Digital is betting its platform, which allows AI agents to move across organizational and classification boundaries without pooling sensitive data, can resolve this contradiction at scale.
- Ryland's hire — fourteen years at AWS including founding its Office of the CISO, thirteen years at Microsoft in global standards roles — signals that Istari is treating this as an infrastructure problem requiring deep institutional experience, not a product problem requiring faster iteration.
- The company now faces the harder test: proving that its decentralized architecture can deliver speed and security simultaneously in environments where failure carries consequences far beyond the commercial.
Mark Ryland has spent three decades building systems for organizations that cannot afford to fail. His appointment as Vice President of Artificial Intelligence at Istari Digital, announced in late May, represents a deliberate wager by the Arlington-based startup on the kind of expertise required to solve a defining problem in defense and aerospace: how to deploy AI agents across security boundaries without consolidating the sensitive data those boundaries exist to protect.
Ryland's career has been shaped by exactly this tension. At Amazon Web Services, he spent fourteen years first building the public sector business as Chief Solutions Architect, then founding and directing the Office of the CISO — the body responsible for security strategy across the entire company. Before that, he held senior roles at Microsoft for thirteen years, including Global Head of Standards. The consistent thread: making trustworthy infrastructure work in environments where trust is not optional.
The problem Istari is trying to solve is not about building smarter agents. It is about making agents safe to operate when they cannot see all the data they might need, must respect organizational boundaries, and must remain accountable to stakeholders who cannot easily share information with one another. CEO Will Roper framed the challenge plainly: every company in defense and aerospace is embracing AI, but the hard problem is making agents safe to operate across security and organizational divides.
Istari's platform is designed to let data and AI agents move across tool, organizational, and classification boundaries without requiring the data consolidation that would violate security protocols. Whether the company can deliver on that promise at scale — in environments where failure carries consequences far beyond a lost customer — is the question Ryland's appointment is meant to help answer.
Mark Ryland has spent the better part of three decades building infrastructure for organizations that cannot afford to fail. Now he's moving to a company built around a single conviction: that the next generation of artificial intelligence will live or die based on whether it can operate securely across the boundaries that keep sensitive work separate.
Ryland's appointment as Vice President of Artificial Intelligence at Istari Digital, announced in late May, represents a significant bet by the Arlington-based startup on the expertise required to solve what may be the defining technical problem in defense and aerospace right now. These industries are moving toward AI agents—autonomous systems that can coordinate work across teams and organizations—but they face a constraint that consumer tech companies do not: they cannot simply pool their data in one place. Security classifications, intellectual property concerns, and organizational autonomy all demand that sensitive information stay compartmentalized even as AI systems need to operate across those boundaries.
Ryland arrives with the kind of pedigree that signals Istari's ambitions. He spent fourteen years at Amazon Web Services, first building out the company's public sector business as Chief Solutions Architect, then spending eight years founding and directing AWS's Office of the CISO—the office responsible for security strategy across the entire company. Before that, he held senior roles at Microsoft for thirteen years, including positions as Global Head of Standards and U.S. National Standards Officer. His career has been defined by one consistent thread: making trustworthy systems work in environments where trust is not optional.
The problem Istari is trying to solve is not about making smarter AI agents. The industry has plenty of smart agents. The problem is making them safe to operate when they cannot see all the data they might need, when they must respect organizational boundaries, and when they must maintain accountability to multiple stakeholders who cannot easily share information with one another. It is, in other words, an infrastructure problem—the kind of problem that takes years to solve and requires someone who understands both the technical constraints and the organizational politics of high-stakes environments.
Will Roper, Istari's CEO, framed the hire as a turning point. "Every company in defense and aerospace is now embracing AI," he said, "but the hard problem isn't building smarter agents—it's making them safe to operate across security boundaries and across organizations that can't easily share data." Ryland's appointment, in this framing, is not about hiring a talented executive. It is about bringing in someone who has already spent a career solving exactly this kind of problem at scale.
Ryland himself described the shift underway in AI as a move from individual tools to connected systems of work. That shift creates opportunity, he suggested, but only if organizations can preserve trust, control, and accountability as AI begins to operate in real mission environments. Istari's decentralized platform is designed to let data and AI agents move across tool, organizational, and classification boundaries without requiring the kind of data consolidation that would violate security protocols or organizational autonomy. In theory, this means defense and aerospace programs can move faster while preserving the security and control that their missions demand.
What remains to be seen is whether Istari can deliver on that promise at scale. The company is entering a market where the stakes are genuinely high—where a failure in security or accountability could have consequences far beyond a lost customer or a damaged reputation. Ryland's three decades of experience in exactly these environments suggests the company is serious about the technical and organizational challenges ahead.
Citações Notáveis
Every company in defense and aerospace is now embracing AI, but the hard problem isn't building smarter agents—it's making them safe to operate across security boundaries and across organizations that can't easily share data.— Will Roper, CEO of Istari Digital
AI is moving from individual tools to connected systems of work. That agentic shift creates enormous opportunity, but only if organizations can preserve trust, control, and accountability as AI begins to operate across real mission environments.— Mark Ryland
A Conversa do Hearth Outra perspectiva sobre a história
Why does a company building AI infrastructure for defense need someone with Mark Ryland's background specifically?
Because the problem isn't the AI itself. It's that defense and aerospace organizations can't consolidate their data the way a tech company can. Ryland spent his career at AWS and Microsoft building systems that work across those kinds of boundaries—where trust and security aren't afterthoughts, they're the foundation.
So this is really about security architecture, not AI innovation?
It's both, but security comes first. You can have the smartest AI agent in the world, but if it can't operate across classified networks or organizational silos without exposing sensitive data, it's useless in a defense context. Ryland understands that constraint from the inside.
What does "agentic AI" mean in this context?
It means AI systems that can act autonomously—coordinate work, make decisions, move information—across multiple teams and organizations. The challenge is doing that without centralizing all the data in one place, which would violate security protocols.
Is this a solved problem, or is Istari betting on something that might not work?
It's unsolved. That's why Ryland's hire matters. He's not coming in to execute a known playbook. He's coming in because he's spent decades building trustworthy infrastructure in environments where failure isn't an option. That's the kind of thinking this problem requires.
What happens if they succeed?
Defense and aerospace programs move faster without sacrificing security. Teams collaborate across fragmented ecosystems while maintaining control of their own systems and intellectual property. It's the difference between speed and security being a tradeoff versus being compatible.