CDAO Fall schedule

• Data infrastructure investment is notoriously difficult to cost-justify in traditional financial frameworks because the value is cumulative, diffuse, and often realized by business units that didn't fund it. How are practitioners solving this problem?
• CFOs are increasingly being asked to approve significant AI and data platform spending with ROI projections that are genuinely hard to validate. What does a credible, honest investment case look like?
• How do you have an honest conversation with finance leadership about the fact that some of the most valuable data investments won't show a return for two or three years?
• What metrics have you found actually resonate with finance leadership, beyond the ones that data teams naturally reach for?

Panelists:

Steve Boras, Head Model Risk Management & Validation - Citizens Financial Group

Dartanion Williams, Chief Data Officer and Executive Director – CITY OF BALTIMORE

Dr. Tarun Sood, Chief Data and AI Officer, AMERICAN CENTURY INVESTMENTS

Justin Heller, Former SVP, Chief Data Officer, SYNCHRONY FINANCIAL

Moderator: David Blaszkowsky, Senior Program Manager – EDM Association

• Most organisations have a compelling AI strategy and a frustrating implementation reality. What are the structural, cultural, and operational factors that determine whether a data and AI programme scales or stalls?
• Scaling AI beyond a handful of use cases requires standardized platforms, shared infrastructure, and consistent practices across business units that often have very different priorities. How do you build that without becoming the team that slows everyone down?
• Where does centralization help and where does it hurt? The organisations furthest ahead seem to have found a specific balance between central governance and domain autonomy. What does that balance look like in practice?
• Executive sponsorship is easy to get at the start of a transformation and notoriously difficult to sustain through the messy middle. How do data leaders keep leadership genuinely invested when the results are still building?
• What does "scaled" look like as an end state, and how do you know when you have genuinely got there versus just having more activity?

Panelists:

Moshmi Sanagavarapu, SVP Data Management, OMNICOM

Alexander Shvets, Head Medical & R&D Data Science & AI – IPSEN

Michael Mangione, Global Head - Digital & AI Experience Transformation - W. L. GORE & ASSOCIATES

Anu Sundaram, Vice President Business Analytics – RUE GILT GROUPE

Moderator: Haroon Abbu, Senior Vice President, Digital Technology, AI & Data Analytics & CDAIO– BELL AND HOWELL

• Which data quality challenges pose the greatest risk to AI success?
• Where do data silos limit visibility, insights, or innovation?
• Who owns accountability for AI-related decisions and outcomes?
• How important is explainability for business-critical AI decisions?

AI is expensive, operationally complex, and increasingly impossible to justify as a blanket investment strategy. As organisations move from experimentation to scale, the real constraint is no longer ambition, it's prioritization, cost discipline, and deciding what not to do.

• If AI is not cheap to build or run, what use cases should simply be stopped, even if they look innovative?

• Are organisations being honest about ROI, or are they subsidizing experimentation under the label of transformation?

• Who should have the authority to shut down AI initiatives that don't deliver measurable impact — the CDAO, CAIO, or the business?

Amin Assareh, Vice President, Data Science, FIDELITY INVESTMENTS 

Mitch Holt, AVP – Head of Data & Analytics, Pet Insurance – METLIFE

Santiago Ibanez, GENAI Lead – MAPFRE

Moderator: David Woods, SVP – Precisely Global Services – PRECISELY

• Optimizing cost structures by balancing compute intensity, model complexity, and usage patterns

• Prioritizing performance through rigorous evaluation of accuracy, latency, and scalability

• Designing autonomous systems that can act independently while respecting defined boundaries

• Establishing control mechanisms to govern agent behavior, decision rights, and escalation paths

• How has the definition of “accountability in AI” evolved in the past few years?
• What does “success” look like for AI beyond model accuracy?
• What leading indicators should organizations track to ensure AI initiatives are on the right path?
• How are organizations preparing for evolving regulatory requirements around AI?

Panelists:

Ramesh Natarajan, Head - GenAI/ML Solutions, DAIICHI SANKYO US

Prashant Singh Tewatia, Vice President, JPMORGAN CHASE & CO

Jordi Rodriguez Oliver, Head of CoE (Center of Excellence), MABE GLOBAL

Andrew Ekpenyong, Director, Data Program Management, CITI

Celio Oliveira, Chief Data and AI Officer, GOVERNMENT OF CANADA

Moderator: Haroon Abbu, Senior Vice President, Digital Technology, AI & Data Analytics & CDAIO, BELL AND HOWELL

• What does a successful transition from AI vision to measurable business value look like?
• Which business objectives should be prioritized when defining an AI strategy?
• How can organizations ensure alignment between business strategies, technology investments, and data capabilities?
• What are the key enablers of successful end-to-end AI transformation?

Join us on Day 2 for an informal networking lunch designed to bring together peers, speakers, and sponsors in a relaxed setting away from the exhibition hall, while still keeping you close to the energy of the event.

Use this time to pick up conversations from the morning sessions, share ideas with fellow data and AI leaders, and meet new contacts from across the community.

There will be no presentations and no pitches, just lunch, conversation, and the opportunity to build valuable connections.

In this session, you will learn to:

• Identify workflow gaps: Discover the two key gaps that must be closed to give business data users the same AI-driven revolution that coders already enjoy.
• Streamline data workflows: Learn how an inline AI chat experience can optimize data onboarding, preparation, and the publication of data products.
• Enhance reliability: Understand how to build seamless data processes that allow you to easily check, rerun, and share your results.
• Leverage key technologies: Explore how to use ontologies, memory, and verifiability to anchor AI data work in trusted, repeatable systems.

• Trust in AI systems is eroding at roughly the same speed that AI adoption is accelerating. What is driving that erosion and what are organisations doing about it that is actually working?
• The employees who are closest to AI systems are often the most skeptical about them, because they see failure modes that leadership does not. How do you build internal trust with the people who know enough to be legitimately critical?
• Customer trust in AI-driven products and services is fragile and asymmetric. It takes a long time to build and can be destroyed by a single high-profile failure. How are organisations managing that risk structurally rather than reactively?
• Transparency about how AI systems work is increasingly expected by regulators, customers, and employees. How much transparency is genuinely achievable with complex models, and where does the honest answer fall short of what people want to hear?

 AI is no longer just a technology discussion—it is a business, governance, and leadership challenge. This keynote examines how boards and executive teams can move beyond experimentation to create measurable business value while maintaining trust, compliance, and accountability. Through real-world examples and practical frameworks, attendees will learn how to govern AI responsibly, prioritize high-impact opportunities, and position their organizations for long-term success in the AI era.  

Financial services teams need actionable answers. Delivering them just in time improves the experience for internal and external clients. What changed? Where is the exposure? Which clients are affected? Finding out often means piecing together data across disconnected systems and tracking down the people who know what it means.

Drawing on real customer examples, this session explores how PromptQL connects enterprise data with the expertise needed to interpret it. We will examine a major U.S. bank building governed AI access to enterprise data, a global investment bank using AI to investigate settlement failures and a global payments company using AI to answer fraud investigation alerts. These examples will anchor a broader discussion of lessons for risk analysis, reporting, research and client service across financial services.

We will examine what makes these workflows effective: connected data, expert knowledge and clear access controls. We will also discuss how to keep people accountable for decisions. Attendees will leave with practical lessons for choosing a high-value AI workflow and evaluating the quality of its answers.  

• Synthetic identity fraud, deepfake-assisted account takeover, and AI-generated social engineering are accelerating at a pace that is outrunning most institutions' detection capabilities. What does the current threat landscape actually look like for a tier one financial institution in 2026?
• Real-time fraud detection requires infrastructure and decisioning engines operating in milliseconds without false positive rates that destroy the customer's experience. What does the architecture that makes this work actually look like in production?
• The arms race dynamic means improvements in defensive AI tend to be matched relatively quickly by adversaries who have access to the same foundational technology. Where is the sustainable edge for financial institutions, and does it even exist?
• How are firms thinking about the liability and regulatory implications when an AI fraud detection system makes a consequential error, either missing genuine fraud or blocking a legitimate customer?

• Unlocking AI-driven growth while maintaining strategic and operational control
• Establishing governance and risk frameworks that enable innovation at scale
• Managing security, compliance, and regulatory challenges in a rapidly evolving landscape
• Building trust through responsible AI practices and transparent decision-making

Key Points:

• Ownership is a strategy decision, not a governance artifact: which few decisions carry the value, and which business leader carries the consequence when each is wrong. Settle both before the money moves. Which of your current AI bets could name them today?
• Diffused accountability is an operating model: every data and AI outcome has sponsors and none has an owner, because owning it puts one name on the hook for being wrong. That is a design, not an accident. Why does your organization keep choosing it?
• AI did not create the gap. It removed the people covering for it and now acts on it at machine speed: two teams report different numbers, both right, and the agent picks one. Whose review moves when it has been wrong for six months?
• One question in every funding meeting: whose objectives carry this outcome? Data leaders who ask it push the decision back to the executive who claimed it. Those who do not absorb accountability for outcomes they were never given. Which one are you?

Discussion points:

• How to prioritize AI use cases across the enterprise without chasing every new capability
• Where AI adds genuine value versus where traditional automation may be faster, safer, or more cost-effective
• How to turn AI strategy into execution across teams, processes, and business units
• What leaders need to consider when balancing innovation, risk, governance, and operational readiness

• The pipeline of future data and AI leadership talent is real but incomplete. What are the gaps between how organizations are developing the next generation of data leaders and what those leaders are actually going to need?
• Technical excellence gets people into senior data roles. It is rarely what determines whether they succeed once they get there. What are the non-technical capabilities that make the difference?
• Mentorship, sponsorship, and deliberate career architecture for data talent are all underinvested relative to their impact. What do organizations that do this well look like?
• What is the one thing you wish someone had told you earlier in your career that would have meaningfully changed your trajectory as a data leader?

• The CDAO and CAIO roles were often created independently, sometimes in competition with each other, and occasionally in confusion about where one ends, and the other begins. Now that both functions are maturing, what have you learned about how to make this partnership work in practice?
• The data layer and the AI layer are inseparable in theory but frequently siloed in reality. Where does the friction between these two functions live day to day, and what does it cost the organization when it goes unresolved?
• There is a version of this relationship where the CDAO and CAIO spend most of their time negotiating with each other over ownership, budget, and priorities. And there is a version where they operate as a genuine leadership unit that is stronger than either function alone. What is the difference between those two organisations and how do you deliberately build the second one?
• If you could say one honest thing to the person sitting across the aisle in the other community tonight, something you would not normally say in a formal setting, what would it be?

Adem Albayrak, Chief Data and AI Officer – ALZHEON

Jillian Landi, Chief AI Officer, NEEDHAM BANK

Anitha Sivasubramanian, Chief AI Officer, LYCA GROUP

Moderator: Jordi Rodriguez Oliver, Head of CoE (Center of Excellence), MABE GLOBAL

• Managing the evolution of the CAIO role from technology advocate to business strategist
• Preparing organizations for an AI-first future through leadership, governance, and operating-model transformation
• Evaluating whether today's CAIOs are ready to lead the next phase of enterprise AI transformation
  
  

• Human in the loop has become a reassuring but often meaningless shorthand for AI oversight. When systems are making thousands of decisions per second, what does meaningful oversight actually look like?
• The cognitive load of overseeing AI systems is a growing problem organisations are only beginning to grapple with. How do you design oversight that humans can actually sustain at scale?
• Different decisions warrant different positions on the spectrum from full human control to full autonomy. How are organisations making those calibration decisions systematically rather than case by case?
• The most dangerous moment in AI deployment is often not the launch, when everyone is watching, but six months later when oversight becomes routine and then perfunctory. How do you design against that?