Zero Trust, Real Networks, and the AI Debt Collector— with JJ Jabbusch
In this episode of the No Trust Podcast, Jaye Tillson and John Spiegel welcomed JJ Jabbusch back to the show for a wide-ranging conversation about Zero Trust, network security, technical debt, agentic AI, and the growing complexity of enterprise networking.
PODCAST
John Spiegel
11/20/20269 min read


Why Zero Trust, network architecture, automation, and the next generation of network engineers matter more than ever.
In this episode of the No Trust Podcast, Jaye Tillson and John Spiegel welcomed JJ Jabbusch back to the show for a wide-ranging conversation about Zero Trust, network security, technical debt, agentic AI, and the growing complexity of enterprise networking.
JJ is a longtime networking and cybersecurity professional, consultant, and community advocate who is also launching NetSecConf, a vendor-neutral network security conference planned for North Carolina in 2027.
What started as a discussion about why network security teams struggle to implement Zero Trust quickly turned into something much bigger: Have we made enterprise networking so complicated that we're struggling to secure the infrastructure we've already built?
And what happens when AI agents begin navigating those environments faster than humans can understand, troubleshoot, and repair them?
The conversation explored why network and security teams often work against each other, how undocumented configurations become technical debt, why automation alone won't solve architectural problems, and what the next generation of network engineers needs to learn.
Listen Here - https://on.soundcloud.com/gon6HaJ5TExWxbDHxM
The Real Zero Trust Problem Isn't Always Technology
Zero Trust is often presented as a technology challenge. Deploy identity controls, implement segmentation, introduce network access control, and continuously verify access.
But anyone who has actually attempted a significant Zero Trust transformation knows the reality is more complicated.
John opened the discussion by pointing to one of the most persistent barriers: the disconnect between networking and security teams.
Security teams define objectives, often without fully understanding the technical complexity of implementing them. Network teams are then asked to make changes that may sound simple but involve multiple vendors, management platforms, access policies, and legacy systems.
JJ argued that part of the problem starts with the language itself.
Different teams may have very different interpretations of Zero Trust. They also bring different priorities and different understandings of their environments.
Security wants to reduce risk. Networking needs to maintain connectivity, performance, and availability.
Neither objective is wrong.
The problem arises when organizations fail to establish a common understanding of what they're actually trying to accomplish.
JJ emphasized that networking teams need clear security objectives rather than highly prescriptive instructions from people who may not understand the underlying infrastructure.
Once those objectives are established, engineers need the freedom, knowledge, and resources to design an appropriate solution.
The takeaway: Successful Zero Trust projects require shared outcomes and architectural understanding, not simply instructions to configure more security features.
We've Become Better at Configuring Networks Than Architecting Them
One of JJ's strongest observations concerned the growing gap between configuring individual devices and understanding the network as a complete system.
Enterprise environments are increasingly complicated.
Different vendors implement segmentation differently. Operating-system versions introduce their own exceptions. Identity mechanisms vary across devices. Some endpoints support certificates; others can only be identified through MAC addresses. OT and IoT environments introduce additional limitations.
Now add firewalls, NAC, SASE, SSE, ZTNA, cloud networks, and years of configuration changes.
The result can be an environment that works but is extraordinarily difficult to understand.
JJ argued that organizations have become too focused on configuring individual components rather than architecting systems that can evolve safely.
Documentation is frequently incomplete. Rules reference objects that nobody remembers creating.
Temporary fixes become permanent. Engineers inherit environments without understanding why particular decisions were made.
John shared an example from a large healthcare organization where the security team wanted to block just two protocols.
Simple enough, right?
Not quite.
The organization operated equipment from three networking vendors, each with different segmentation mechanisms. Introducing identity-based controls and an existing ZTNA platform added further complexity.
What sounded like a straightforward security requirement became a significant architectural exercise.
The takeaway: You cannot manage the security of a network effectively if you don't understand how its components interact.
Agentic AI May Become the Next Collector of Technical Debt
Perhaps the most interesting part of the discussion came when John introduced a provocative idea.
Historically, ransomware has been the event that forces organizations to confront years of accumulated technical debt.
After an incident, teams discover that a NAC deployment was never completed, a segmentation initiative stalled, old remote-access services remain exposed, or firewall exceptions have survived long past their original purpose.
The organization suddenly finds the budget and urgency to address problems that existed for years.
John suggested that agentic AI may become the next technical debt collector.
Why?
Because AI-assisted systems could increasingly discover, correlate, and exploit weaknesses far faster than traditional human-driven processes.
And those systems don't necessarily have to be controlled by external attackers.
An internally developed AI agent with excessive privileges, poorly understood tool access, or an objective that produces unintended consequences could create its own security problems.
JJ agreed with the broader argument while cautioning against unnecessary AI fearmongering.
The important issue isn't whether every frightening prediction about AI comes true.
It's that organizations are already struggling with complexity, and increasingly capable automation may make the consequences of that complexity emerge faster.
This brings the discussion back to Zero Trust.
If an attacker—or an autonomous agent—obtains access to one part of the environment, what can it reach next?
Can it communicate with unrelated systems? Access sensitive applications? Traverse management networks? Interact with OT devices?
The number of security products deployed tells us relatively little about the answers.
The takeaway: The security metric that matters increasingly is blast radius—how much of the environment a compromised identity, device, or agent can reach.
Why Temporary Network Fixes Become Permanent Security Problems
Jaye shared an experience familiar to almost every network engineer.
His team implemented access control lists to improve segmentation. The changes unintentionally disrupted an entire factory.
The immediate response was obvious: roll back the ACLs and restore operations.
But the experience created so much concern about another outage that it took six months before anyone was willing to revisit the change, followed by another month of testing.
It's a story that illustrates why technical debt accumulates.
When a business-critical system fails, restoring service becomes the priority.
An engineer may disable a rule, widen a permission, or introduce an exception to get the environment running again.
The immediate problem disappears.
But the temporary configuration may remain in place for years.
JJ argued that this is precisely why architecture and documentation matter.
Understanding why something works is just as important as understanding why it fails.
When organizations understand their dependencies and traffic flows, they can introduce security controls with greater confidence.
When they don't, every change becomes a potential production incident.
The takeaway: Security controls that cannot be implemented and maintained without fear of disruption are unlikely to deliver their intended value.
Can Network Automation Help Us Escape the Complexity?
The conversation then turned toward a possible solution: network automation.
John asked whether technologies such as MCP, automation gateways, Python, and NetOps could help organizations establish a more consistent view of heterogeneous networks and reduce the burden of vendor-specific management.
JJ was enthusiastic about the opportunity but raised an important concern.
We've been here before.
Software-defined networking promised to simplify network operations and make infrastructure more programmable. Yet interoperability challenges, proprietary implementations, and operational complexity prevented many early visions from becoming reality.
Network automation faces some of those same challenges today.
An organization can build powerful scripts and integrations, but somebody must understand the underlying architecture, maintain those tools, and ensure they're safe to execute.
And not every organization has the personnel or resources to develop that capability.
John suggested that AI could help translate configurations between vendors, validate firewall policies, and identify rules that became overly permissive during emergency changes.
JJ agreed, particularly when AI can combine configuration analysis with vulnerability and operational context.
But she also emphasized an important distinction: someone still needs to define the intended outcome.
AI may help interpret configurations and translate policies. That doesn't mean it should independently determine what the enterprise's security architecture ought to be.
The takeaway: Automation can reduce operational complexity, but it cannot replace architectural intent, validation, and accountability.
Network Engineering Needs a New Generation of Skills
Jaye raised another concern: Are we doing enough to attract younger engineers into networking?
Cybersecurity receives enormous attention, while networking can appear less exciting—even though virtually every modern application, cloud service, and security control depends on connectivity.
JJ challenged the notion that networking is any less interesting or important than cybersecurity.
In fact, she argued that effective security architecture requires a strong understanding of the underlying network.
Her concern was that much of today's training is too vendor-specific.
Engineers may become highly proficient with one platform without developing the foundational knowledge needed to understand protocols, architecture, and interoperability across different environments.
JJ described how her early networking education emphasized Layer 2 and Layer 3 fundamentals, industry standards, and integration between competing vendors.
Those principles transferred across the many platforms she subsequently worked with.
The discussion also explored a possible evolution of the network engineering role.
Instead of spending most of their time configuring individual devices through command-line interfaces, future engineers may increasingly use automation, AI-assisted analysis, and policy validation to manage networks at a higher level.
But the fundamentals remain essential.
Someone still needs to understand what the network is supposed to do and recognize when automation gets it wrong.
The takeaway: The future network engineer may use AI and automation more extensively, but success will still depend on vendor-neutral technical foundations and architectural thinking.
JJ's Advice: Start With One Hour a Week
As the conversation approached its conclusion, Jaye asked JJ what network engineers should do tomorrow to prepare for the changes ahead.
Her answer was refreshingly practical.
Set aside time to learn.
It doesn't have to be a formal training program or a major certification effort.
Even one hour a week can make a difference.
Start with something relevant to your current work. Read a technical paper. Explore a new automation concept. Listen to a podcast. Learn what Python can do for network operations. Experiment with approved AI tools in a safe environment.
You don't need to become an expert in everything immediately.
The important thing is to build momentum.
JJ suggested that after only a few learning sessions, engineers may discover a technique that could simplify a recurring task or solve a problem they've been living with for years.
That is how organizations begin escaping the endless cycle of break-fix work.
And leadership has a role to play.
If engineers spend every available hour keeping the lights on, they will never have the opportunity to improve the systems creating that workload.
Organizations must create space for learning, architectural thinking, and operational improvement.
Bringing the Network Security Community Back Together
JJ also discussed her plans for NetSecConf, a vendor-neutral network security event being developed for North Carolina. Inspired by conferences such as Interop, the objective is to bring networking and security professionals together across vendor boundaries.
Rather than centering discussions on individual products, the event aims to emphasize practical learning, interoperability, emerging technologies, and the future of networking.
It's an idea that connects directly to the broader episode.
If networking and security teams are going to solve increasingly complicated problems, they need opportunities to share knowledge and learn from people outside their usual vendor and organizational environments.
And, Of Course, Thanksgiving
No episode of No Trust would be complete without a conversation about food.
Jaye departed from the usual pineapple-on-pizza debate to ask JJ about American Thanksgiving traditions.
JJ explained that while turkey would be on the menu for family reasons, she would much rather have ribs, barbecue, or sausage alongside mashed potatoes and her grandmother's gravy recipe.
That led to a spirited discussion about Yorkshire puddings, Yorkshire pudding wraps, and whether the traditional British roast dinner might actually improve the American Thanksgiving experience.
John then shared his family's plan for a Thanksgiving version of Hot Ones, involving progressively hotter sauces and increasingly uncomfortable questions.
Whether that becomes a future podcast episode remains to be seen.
Final Thoughts: We Need to Make Networking Understandable Again
The central message from this episode is that enterprise networking has become too complicated to manage through configuration changes, vendor-specific tools, and institutional memory alone.
Zero Trust requires architectural clarity.
Segmentation requires an understanding of legitimate communications.
Automation requires a clearly defined objective.
And agentic AI raises the stakes because it may accelerate both the discovery of weaknesses and the consequences of poorly controlled actions.
The solution isn't another product promising to make complexity disappear.
It's a combination of stronger network fundamentals, better documentation, shared security objectives, cross-vendor integration, thoughtful automation, and investment in the engineers responsible for operating these environments.
Perhaps the most important question organizations should ask is this:
If an AI-powered attacker—or one of our own autonomous agents—begins exploring our infrastructure tomorrow, do we understand our network well enough to predict and limit where it can go?
If the answer is no, that may be the most important technical debt to address.
Listen to the full episode of the No Trust Podcast featuring JJ Jabbusch, hosted by Jaye Tillson and John Spiegel. - https://on.soundcloud.com/gon6HaJ5TExWxbDHxM
Also, make sure to follow the Zero Trust Forum on LinkedIn and sign up for the newsletter!
About the No Trust Podcast
The No Trust Podcast, hosted by Jaye Tillson and John Spiegel, explores Zero Trust, cybersecurity, networking, AI, and the real-world challenges of securing modern enterprise environments. Through conversations with practitioners, technologists, and industry leaders, the show focuses on moving beyond marketing claims to understand what actually works
