Reflections from Red Hat Summit and ONUG AI Networking Summit
By Jeff Gray, CEO & Co-Founder, GluwareÂ
I left two conferences this month, Red Hat Summit in Atlanta and the ONUG AI Networking Summit in Dallas, with a sharper version of something I’ve believed for years: the race to agentic network operations is moving faster than the trust infrastructure required to make it safe. That gap is the most important problem in enterprise networking right now, and it connects everything I heard at both events.
The Threat Environment Is Outrunning Human Response
Security conversations at ONUG centered on what participants called the “patch storm.” AI is now accelerating vulnerability discovery so rapidly that quarterly remediation cycles are already obsolete. In fact, the common wisdom is that the 30-60-90 day patch window is going away, with the 24 hour patch window taking its place. The fear is tangible.
So, awareness is no longer the gap. The real problem is trusted execution speed – the ability to detect a vulnerability, map it to live network configurations, execute immediate compensating controls and close the loop with validated remediation before the exposure window becomes a breach. That is what Gluware Titan Exposure Management does.
ONUG Co-Founder Nick Lippis said it directly when recognizing Titan’s Best in Show win in the Agentic AI category: “Titan maps live network configurations directly to known CVEs and enables validated, closed-loop remediation workflows, helping close the gap between detection and response at a speed and scale security teams can’t match manually. Visibility alone isn’t enough. Response has to keep pace with how fast threats move.”
That’s the right framing — and it’s exactly what we built Titan to deliver.
Agentic Orchestration Without a Trust Layer Is a Liability
Both conferences were enthusiastic about agentic AI — and appropriately cautious about it. The consistent thread at ONUG was that governance and trust are not features you bolt on later. They are the foundation you build first.
Nick Lippis put his finger on the core enterprise challenge ahead of the summit: “Today’s environments are filled with fragmented tools — observability platforms, ITSM systems, security controls — that don’t communicate well. As multiple AI agents are introduced into each of these domains, the lack of coordination becomes a serious problem. There’s no shared understanding of workflows, no consistent control plane, and limited visibility into how agents interact with each other.”
The ONUG main stage keynote framed the stakes plainly: “The rise of AI agents changes how networks are built and secured.” CodiLime’s post-event recap put the operational risk in a more pointed way: “A network agent that gives a wrong recommendation is one problem. A network agent that pushes a wrong change automatically is a much bigger problem.”
I’ve said for years that every enterprise network is a snowflake — heterogeneous, brownfield, and stubbornly multi-vendor. Generic agentic frameworks can reason and recommend, but they cannot safely execute across 56 vendor OS platforms without a validated semantic translation layer. That is not a gap you can architect around with an orchestrator. And it’s not something that you can just trust an agent to figure out by trial and costly error. It has to be built, device by device, OS by OS, over years.
That is why Gluware invested over one million engineering hours in DIAL (our Device Interaction Automation Layer), validated across 306 releases and scale tested to 450,000 network devices. We built the trust layer before building Titan AI. Every agentic action Titan takes runs through DIAL for semantic translation and outcome validation. Trust infrastructure first, agentic workflows second: that is the only responsible order of operations, and it is a sequence most of our competitors have skipped.
The Network Needs a Specialist, Not Just an Orchestrator
ONUG has consistently argued that AI networking is a cross-domain discipline, not a single-platform problem. Nick Lippis wrote ahead of the summit that enterprises need to focus on “connecting AI agents across multi-vendor environments, normalizing workflows between IT and business systems, and building a control plane for agentic operations.”
Red Hat Summit showed that Ansible Automation Platform is maturing into the enterprise execution layer for agentic IT operations — AAP 2.7 is designed to operationalize AI agents at enterprise scale. That is genuinely powerful. In-depth specialization to ensure valid automation of enterprise networks complements that broad reach. Again, the reason is that enterprise network environments have accumulated decades of multi-vendor complexity that no horizontal orchestrator was built to understand at the semantic level.
Trying to automate a complex, multi-vendor network with a broad orchestration layer alone is like doing precision surgery with a Swiss Army knife: capable instrument, but wrong level of precision. Gluware is tailored to be the specialist layer for network automation: semantic translation, multi-vendor validation, and governed agentic execution across 56+ OS platforms. Ansible handles the broad IT automation surface; Gluware handles the network depth. Together they form an architecture program, which is what ONUG said enterprises actually need to build.
The Trust Layer Is the Linchpin
The AI era is not waiting. Threats are faster, infrastructure is more complex, and an unvalidated agentic change at scale is hard to reverse. What both summits confirmed is that the defining variable is not which agentic platform you choose; it is whether the trust layer beneath it can actually be relied upon.
That is what Gluware was built for. If you’re ready to put trusted agentic execution into practice, request a demo and see what Titan AI and Titan Exposure Management can do for your network.
