Governing the Invisible
Enterprise AI Beyond the Sandbox
Most enterprises have AI in production. Few can explain how it works, prove it is safe, or govern what they cannot see. This panel brings together leaders from CrowdStrike, ServiceNow, Itron, MNP, and Adient to discuss how ontology-driven architectures make enterprise AI traceable, auditable, and governable at scale.
Recorded
June 10, 2026
Duration
60 Minutes
Format
Virtual Panel
60-minute panel — watch the full recording below
Key Takeaways
Honest Answers to the Hard Questions
Why 77% of enterprises have updated AI security strategies but only 26% can enforce them in real-time
How to govern decentralized AI when three-quarters of employees use unsanctioned tools your security team cannot see
The structural difference between sovereign AI and secure AI, and why private infrastructure alone does not solve governance
A practical zone-based governance framework that matches controls to user personas instead of one-size-fits-all policy
Leaders Who Joined From
The Models Are Ready. Most Organizations Are Not.
The phase of comfortable AI experimentation is officially over. For modern enterprises, the challenge has dramatically shifted from accessing models to governing, securing, and operating them across highly fragmented ecosystems.
77%
of organizations have updated their AI security strategies
26%
report having the actual architecture to enforce those policies in real-time
Source: Check Point 2026 AI Security Report
The Panel
Meet the Speakers
Leaders from across AI, cybersecurity, data architecture, and enterprise transformation who shared production-grade insights, not theory.
Jake McAndrew
COO
Cyberhill Partners
Patrick McGarry
Federal Chief Data Officer
ServiceNow
Andrew Buffone
Director, Advanced Analytics
MNP Digital
Anant Raigaga
Director, IT Enterprise Architect
Itron
Richard Doak
Director, Innovation & Transformation
Adient
Mark Murtagh
Global Vice President, Solution Architects
CrowdStrike
What Was Covered
Five Hard Questions. One Real Conversation.
No slides. No sales pitches. Five practitioners tackled the structural, technical, and operational problems that every enterprise hits when AI moves from sandbox to production.
The "Minimum Viable Governance" Threshold
Companies often stall because they try to build a perfect, monolithic governance policy before deploying anything. What is the leanest, non-negotiable framework an enterprise needs on day one to safely move a model from a sandbox into production?
Budget and ROI Realities
Industry data shows enterprise AI spending is surging, but the actual number of individual user licenses is shrinking. This means organizations are moving away from mass enablement and toward targeted, high-value systems. How do you structure a governance model that measures the financial value/ROI of a model against its compliance risk?
Decentralized Innovation vs. Centralized Control
Business units (like HR or Marketing) are impatient and are buying "embedded AI" features buried inside standard SaaS tools, bypassing central IT. How do you enforce enterprise-wide compliance when the AI footprint is decentralized and practically invisible?
Sovereign AI vs. Hyperscaler Lock-In
Many heavily regulated global enterprises are terrified of losing control of their intellectual property to major public cloud providers. How are you evaluating the rise of "Sovereign AI" architectures, private AI clouds, or localized data factories to keep data strictly inside your perimeter?
Edge AI and Fleets
When you push AI out to the edge -- whether that is factory floor machinery or distributed smart devices -- you encounter low bandwidth, latency constraints, and physical security risks. How do you manage and govern a fleet of thousands of edge-AI models compared to a centralized cloud model?
Agenda
60-Minute Session Flow
Welcome & Framing
Setting up the problem: enterprise AI is no longer just a model issue. It is a governance, security, and operations issue.
Opening Perspectives
Each panelist shares their view on the biggest blocker to secure enterprise AI adoption from their domain.
Moderated Discussion
Deep dive into the five topics: governance thresholds, ROI realities, decentralized AI, sovereign architectures, and edge fleet management.
Audience Q&A
Audience questions answered live by the panelists.
Closing Takeaways
Each panelist gives one concrete recommendation for enterprise leaders over the next 90 days.