AI is in production. For most organizations, the hard part isn't getting there anymore — it's what comes after.

Once LLM-powered features are live, once GPU-accelerated inference is running at scale, once agentic workflows are touching real engineering and ops processes, the questions change fast. Why is the infrastructure bill still climbing? Why are reliability metrics flat? And why does the team keep finding out something broke from a customer ticket?

This panel is for the people who are past the pilot and actually dealing with those questions.

Rather than walking through a single customer story, we've built a composite scenario most attendees will recognize immediately: a production AI deployment with real inference pipelines, real GPU fleet management, and a board that's starting to ask hard questions about cost and return. Each panelist represents a different layer of that architecture. Together, they'll work through the operational decisions that actually move things forward — not the theory, the execution.

Moderated by [Futurum Analyst Name], this is a practitioner conversation about running AI infrastructure the way your roadmap, your CFO, and your customers all need it to run — at the same time.

The core tension: Most operational models weren't built for this. They were built for stateless workloads, predictable traffic, and human-in-the-loop remediation. That foundation cracks under inference pipelines, agent orchestration, and GPU fleet pressure. The gap between how fast teams are shipping and how well they can observe, govern, and optimize what shipped is where the reliability incidents, cost overruns, and engineering burnout tend to accumulate.

Panelists:

• [Futurum Analyst] — Moderator

• Wayne Seager, RVP of Field CTO and AI Innovation, Dynatrace

• Merlin Yassi, Generative AI Lead, Google — hyperscaler and agentic AI architecture

• [GSI / Ecosystem Partner TBD: ServiceNow · Atlassian · NVIDIA · Dell] — enterprise workflow, infrastructure, or platform perspective

If you're a Director, VP, or C-suite leader accountable for platform reliability, infrastructure efficiency, or the engineering velocity your business depends on — this session will give you a concrete operational framework for running AI workloads the way your organization actually needs them to run: fast, cost-controlled, and reliable enough to stake your SLAs on.