Panel discussion

As AI systems evolve from assistants to autonomous agents, enterprises need a clear operating model, decision rights, governance, delivery standards, and accountability. This panel explores what an effective AI Center of Excellence looks like in practice: who sits on the team, how it partners with business units, how it sets standards for security and compliance, and how it scales successful use cases across the organization.

We’ll discuss what the CoE should centralize vs federate, how to avoid becoming a bottleneck, and how to maintain human oversight for risk management and quality. Attendees will leave with patterns for structuring an AI CoE that accelerates outcomes while reducing risk.

TBC

Panel discussion

UST surveyed enterprise technology decision-makers in industries ranging from financial services and healthcare to manufacturing, retail, and consumer packaged goods to better understand how they are changing, upgrading, and modernizing their cloud and on-premises infrastructure and observability tooling to support AI workloads.

This session will share findings from UST’s survey of enterprise decision-makers and discuss what those shifts mean in real environments, cloud, on-prem, and hybrid. Panelists will unpack where organizations are investing, what’s driving those decisions, and what pitfalls to avoid as AI moves into production.

Presentation

Data modernization and AI are inseparable. AI demands data architectures capable of handling unstructured data, real-time analytics, and continuous compliance monitoring. Traditional batch ETL pipelines and siloed data systems cannot meet these requirements. Without modern, cloud-native, and event-driven data platforms, enterprises cannot achieve the agility, scalability, and trust required for advanced AI use cases.

Modernization enterprise data should center on creating a sentient data stack- an architecture that combines semantic modeling, autonomous agents, and policy-as-code to create adaptive and goal-driven data operations.  To accelerate this transformation, we advise organizations to rethink their data architecture in six core areas: Semantic-first modeling, Agent-ready API design, Retrieval-Augmented Generation/RAG, AgentOps, Responsible AI governance, and human mediation. 

Panel discussion

Success is measured less by novelty and more by repeatable business impact. This panel shares how enterprises quantify value using practical metrics such as Task Completion Rate (TCR), Work Reduction Rate (WRR), and Cost per Task (CPT) and how they translate those measures into funding decisions, operating changes, and scaled deployment.

Leaders will discuss what worked, what didn’t, and why initiatives commonly fall short (data readiness, adoption, process redesign, controls, or unclear ownership).

Panel discussion

With the rise of AI used in the SDLC, IT leaders must rethink application security and governance across infrastructure, applications, and LLMs. Managing risk in the age of ‘white coding’ is an essential requirement for managing risks such as hallucinations, context overload, and model drift.

Many organizations have acquired disparate tools to improve observability. For AI in the SDLC, organizations need an architectural layer that integrates into the engineering platform and feeds development models. An integrated observability layer becomes the nervous system for AI.  

Presentation

AI programs fail as often from unmanaged risk and unclear accountability as from technical limitations. This session outlines a practical approach to Responsible AI that enables scale: establishing cross-functional governance (e.g., an AI Council), defining controls and escalation paths, and creating an incident and measurement loop for continuous improvement.

We’ll discuss how risk mitigation differs by use case (customer-facing vs internal, generative vs predictive, high vs low impact), how to embed governance into delivery workflows, and how to demonstrate progress with audit-ready evidence. You’ll leave with a blueprint for implementing Responsible AI without slowing time-to-value.

Workshop

This working session turns the day's insights into a concrete execution plan. Using a guided workbook, participants will take 1–2 real use cases through three phases, Diagnose & Design, Prototype & Prove, and Operationalize & Scale Planning, to produce a 90-day roadmap with named owners, governance checkpoints, success metrics, and dependency tracking (data, security, change management, delivery capacity).

You’ll also identify your biggest blockers and leave with a reusable framework you can run with your teams to replicate the process across additional use cases.