JPMorgan Chase & Co., a financial giant with over $3.9 trillion in assets and operations in more than 100 countries, is not just a leader in banking, it is at the forefront of innovation in data and AI. The firm is pushing the boundaries of financial services, offering customers smarter, faster, and more personalized experiences.

In this second installment of our three-part conversation, Tiffany Perkins-Munn, Head of Data & Analytics for Connected Commerce, Payments & Digital Wallets at JPMorgan Chase & Co., sits down with Mahesh Kumar, CMO at Acceldata, to explore how large enterprises are tackling two fundamental challenges in the age of generative AI: cross-functional collaboration and enterprise-wide governance.

In part 1, Perkins-Munn reflected on her career journey, the human element of data work, and the real-world potential of generative AI. This segment shifts the focus to the mechanics of organizational change.

Rethinking collaboration: From linear handoffs to agile quads

In a complex enterprise like JPMorgan Chase, aligning diverse teams across data, product, technology, and design is not easy. But with AI now embedded into nearly every part of product development, traditional handoff models are no longer sufficient.

“Historically, there might be some structure in place where the teams pass off the work, a little bit more of a linear process. That has changed over time,” Perkins-Munn says.

To address this, her team uses what she calls a “quad structure,” a framework that brings together four core disciplines: Product, Technology, Design, and Data & Analytics.

“In a collaborative quad approach, each of those areas might be using AI to execute their part of the ultimate deliverable, and then coming together to synthesize the deliverable into a meaningful outcome for the key stakeholder.”

This quad model doesn’t just promote efficiency; it fosters cohesion around a shared vision. But to make it work, she notes, everyone must be aligned on the enterprise’s overarching AI strategy.

“You want to make sure that you have an overarching AI strategy so that people don’t go down a rabbit hole.”

The strategy ensures all AI initiatives are coherent, compliant, and ultimately additive to the business, rather than fragmented experiments working in silos.

Why every team needs to know the AI roadmap

With different teams leveraging AI in unique ways, from summarization and automation to data aggregation, alignment is critical. The quad structure provides a format for shared visibility and accountability.

“The quad structure helps us to collaboratively evaluate what’s happening, who’s doing what, which AI components are at play, what the roadmap looks like, and what’s the timing for bringing all of our pieces together,” Perkins-Munn explains.

This shared roadmap not only clarifies execution timelines but also promotes governance discipline. It forces teams to think beyond their functional bubble and toward the bigger picture.

Governance in the age of agents and citizen builders

As AI capabilities become more accessible, including to non-expert users like citizen data scientists, Perkins-Munn emphasizes the need for governance frameworks that evolve accordingly.

“The stakes are particularly high when it comes to data privacy, bias, governance, transparency, and AI-driven initiatives,” she cautions.

She outlines a multi-pronged approach to responsible AI that includes encryption, secure storage, bias mitigation, and inclusivity within data science teams.

Addressing bias, in particular, requires continuous monitoring and diverse team composition. “You have to foster a culture of inclusivity in data science teams so organizations can better identify and correct biases.”

Transparency, meanwhile, is not just an internal issue — it’s a customer imperative.

“The one piece where we all across industries need to do a lot of work is this communication with the customer. Clear communication will help customers understand how their information is being used.”

Perkins-Munn warns that when AI feels opaque, customers may disengage or opt out. And that, she says, is a loss of opportunity for both the organization and the individual.

Local vs. global governance: Balancing personalization and policy

A recurring theme in Perkins-Munn’s governance philosophy is the need to balance local control, enabling personalization and customer-centric experiences, with global oversight of AI strategy.

“Governance has to be both local and global,” she asserts. “There’s a local element to how you are defining audiences but then there’s also a more global view you have to take.”

For example, local teams may focus on tailoring experiences to specific customer segments, while enterprise leaders must ensure all AI activity ladders up to a cohesive business objective, be it engagement, revenue growth, or operational efficiency.

“Whether it’s as a business line or as a broader capability across the firm, we have to think about governance as bringing those two together.”

Building trust

Ultimately, Perkins-Munn believes the key to scalable, responsible AI lies in trust, both internally among teams and externally with customers: “The goal is to build trust with customers. We have to demonstrate a commitment to ethical AI practices, prioritize data privacy, actively address bias, and maintain transparency.”

At JPMorgan Chase, that trust starts with structure: collaborative execution models, clear AI strategies, and robust governance frameworks.

CDO Magazine appreciates Tiffany Perkins-Munn for sharing her insights with our global community.