Precision CX in Regulated Industries: Navigating AI Governance, Data Architecture, and Enterprise Scale with Comcast VP Shri Nandan

Customer service has emerged as the definitive testing ground for enterprise artificial intelligence, forcing heavily regulated sectors like banking, insurance, and healthcare to deploy complex algorithms directly into the hands of consumers. According to recent findings from the U.S. Government Accountability Office (GAO), these digital touchpoints represent some of the earliest and most visible integrations of customer-facing AI. However, this rapid technological expansion is colliding with a complex web of consumer protection mandates, data fragmentation, and evolving regulatory frameworks, creating an urgent need for precise governance and strategic architectural oversight.

The scope of this deployment is already massive. Financial services data from the Consumer Financial Protection Bureau (CFPB) reveals that all ten of the nation’s largest commercial banks now utilize conversational chatbots to manage customer interactions. In 2022 alone, more than 98 million U.S. consumers interacted with a banking chatbot. Yet, this high volume of automation has not come without friction. The CFPB has repeatedly cautioned that poorly architected conversational agents frequently dispense inaccurate information, fail to recognize when consumers are attempting to exercise vital federal rights, and routinely trap customers in automated loops that prevent them from reaching human representatives.

A parallel dichotomy exists within the healthcare sector. Data from the Office of the National Coordinator for Health IT indicates that 71 percent of U.S. hospitals now deploy predictive AI, with implementations dedicated strictly to scheduling surging from 51 percent to 67 percent in a single year. Despite this rapid uptake, health system administrators remain deeply hesitant. According to a national survey published in the Journal of the American Medical Informatics Association, 77 percent of healthcare leaders cite immature AI tools as their primary barrier to adoption, while 40 percent point directly to regulatory uncertainty.

Regulatory bodies themselves are finding it exceptionally difficult to keep pace with the deployment velocity of commercial AI applications. The GAO recently concluded that federal agencies responsible for overseeing credit unions lack the requisite inspection tools and technological frameworks needed to properly govern modern AI systems. This persistent governance gap leaves a dangerous void between how quickly institutions can launch automated tools and how effectively regulators can monitor their real-world impact.

To examine how enterprises can successfully bridge this divide, Yolandi de Weerdt of Emerj recently sat down with Shri Nandan, Vice President of AI Products and Experiences at Comcast. With a career spanning more than two decades leading digital and AI initiatives across telecommunications, healthcare, financial services, and insurance—including leadership roles at Momentum Financial Services Group, Main Line Health, and MetLife—Nandan offered critical insights into scaling customer experience (CX) in high-stakes environments. Her analysis centers on three foundational pillars: bounding AI operational scopes to secure high-stakes interactions, unifying disparate customer data sources to provide reliable real-time context, and centralizing organizational AI governance to accelerate operational scale safely.

Bounding AI Scope to Secure High-Stakes Interactions

The fundamental premise of enterprise AI optimization often assumes a generic, low-risk corporate setting where conversational errors carry minimal consequences. Nandan argues that this assumption fails immediately when applied to the banking, financial services, and insurance (BFSI) and healthcare sectors. In these environments, the emotional register and legal stakes of a conversation vary wildly depending on the use case. A contact center agent helping a consumer purchase a standard insurance policy is executing a routine financial transaction. Conversely, an agent assisting a patient to navigate an oncology appointment is handling a matter of profound personal sensitivity and clinical consequence.

Therefore, the architectural design of any customer-facing AI system must reflect these distinct risk profiles before a single line of code is ever written. Nandan emphasizes that organizations must conduct an honest, explicit evaluation of what they actually want artificial intelligence to accomplish for the consumer, and crucially, where its operational boundaries must lie.

"When you’re designing your agentic system, there has to be an honest discussion about what it is that you want your AI to do to help the customer," Nandan explained during the conversation. "Is it just scheduling and rescheduling appointments, or is it something fairly simple, like looking at your lab work results? If it’s a little bit more complicated, especially in things like oncology or something more serious than that, how would you expect AI to help the customer? I think it’s important for the organization to understand that there isn’t a lot that AI can do in certain situations, and you need human intervention."

Financial services presents a parallel set of challenges centered around accountability and systemic risk. Nandan highlighted the conceptual appeal of automated AI financial advisors, immediately following the proposition with a sobering question: How can an institution verify that the advice generated is entirely sound, and how can it guarantee that the autonomous agent has comprehensively evaluated every available financial product that could maximize returns for the consumer? Constructing an agent with that level of fiduciary depth is extraordinarily difficult. Furthermore, regulatory mandates compound this complexity exponentially. An autonomous financial agent engineered to comply seamlessly with United States federal regulations may immediately violate statutory requirements in the United Arab Emirates—a critical lesson Nandan learned firsthand during her tenure at MetLife, where compliance frameworks shifted dramatically across international borders.

For digital, CX, and AI leaders, these realities necessitate a deliberate approach to scoping. Nandan outlines three tactical imperatives for managing high-stakes AI integrations:

Classify interaction weight by evaluating emotional, clinical, and financial stakes prior to selecting use cases. Routine appointment scheduling and basic account balance inquiries sit safely at one end of the spectrum, while cancer treatment discussions, sophisticated financial planning, and fraud dispute resolutions sit firmly at the other. The litmus test for leadership is whether the organization can explicitly define the precise utility AI should provide to the customer at any given moment.

Treat human escalation as an essential product feature rather than an emergency fallback. System architects must define the exact handoff mechanics upfront. Artificial intelligence should be strictly utilized to manage bounded, highly repeatable interactions, while human professionals must seamlessly absorb situations defined by high emotional, clinical, legal, or financial complexity.

Design for jurisdiction from inception. Because regulatory regimes vary radically by geography, a monolithic, one-size-fits-all agent design is practically guaranteed to fail across multiple markets. Enterprise governance must thoroughly understand the distinct legal parameters of every operating jurisdiction before the underlying conversational agent is built.

Underpinning all of these tactical measures is the necessity of establishing baseline consumer trust. Nandan contends that any automated decisioning system operating within a highly regulated environment must actively assure the user of their safety before attempting to influence their behavior, present financial options, or build long-term brand loyalty.

"Any technology has to build trust," Nandan noted. "It has to be able to say: you’re dealing with a bot, you’re dealing with AI, you’re dealing with technology, but you’re safe. Your information is safe, and you are in good hands. I think it’s important to build that trust as a foundational capability."

Unified Customer Data for Reliable AI Context

When evaluating the primary obstacles preventing institutions from achieving enterprise-scale AI maturity—whether data silos, regulatory bottlenecks, or organizational readiness—Nandan identifies all three as active barriers, but places special emphasis on data architecture and the underlying cultural behaviors that govern information sharing within large enterprises.

In major financial institutions and healthcare networks, customer data is traditionally fractured across numerous disparate business units and legacy databases. The immediate technical challenge is constructing a unified single source of truth from these fragmented repositories. Only once this foundational consolidation is achieved can engineering teams realistically begin preparing that data for artificial intelligence consumption.

Even after achieving unified, AI-ready data, executive leaders must strategically evaluate where the computational processing—the mechanism that transforms raw data into actionable customer context—actually takes place. Nandan describes this phenomenon as the computational "gravity" of the enterprise architecture. If data processing occurs too far away from the active customer interaction, it inevitably introduces latency, degrades system performance, and exponentially inflates operating costs as AI transaction volumes scale. Consequently, modern enterprise data architecture must be treated as an immediate strategic scaling decision rather than a technical detail left to be resolved post-deployment.

While affirming that clean, high-integrity data is "one hundred percent important" for delivering superior customer experiences, Nandan argues that the most stubborn obstacles are cultural rather than technological.

"The problem with creating good data, creating data with integrity, and creating single sources of truth is more cultural than anything else," Nandan observed. "If you have five different sets of disparate teams owning certain aspects of the data, it’s very difficult for you to say you need to give up that data. So there’s a bit of organizational change that needs to come into place that allows the data team to say: this is the data we have, this is how we all come together, this is how we create a source of truth, this is how we keep it fresh, and this is how we can use it in our decisioning systems and to create context."

This cultural reality dictates that achieving AI-ready customer data is primarily an organizational ownership problem that requires aggressive alignment from the executive suite. Without unified leadership and a cohesive data strategy, institutions inevitably construct artificial intelligence tools on top of fragmented customer profiles, resulting in inconsistent service interactions and compounding customer frustration.

Once a robust data foundation is successfully established, Nandan advises that the primary objective of any AI capability should be solving customer pain points with maximum speed. Integrated data uniquely enables hyper-personalization. Armed with fresh, unified context detailing a customer’s historical journey, an autonomous agent can render tailored experiences in real time. For example, the system can instantly recognize if a customer has repeatedly sought assistance for the same unresolved issue and proactively route them down an expedited remediation path. Ongoing performance evaluations of these agents directly inform the product roadmap: failing features are systematically deprecated, successful capabilities are expanded, and the overall customer experience improves through iterative refinement.

Furthermore, Nandan notes that the governance landscape has fractured in direct alignment with data silos. Oversight frameworks that operated under a single digital governance umbrella fifteen years ago have fragmented into distinct disciplines—data governance, AI governance, and context governance—each requiring dedicated guardrails within a regulated business model.

Centralized AI Governance to Ensure Faster Operating Scale

To achieve an operating model where artificial intelligence genuinely elevates service quality at scale, Nandan advocates for a disciplined, sequential order of operations. The absolute prerequisite is the establishment of a formal AI governance practice; without strict guardrails, decentralized teams inevitably pursue disconnected initiatives, resulting in organizational chaos. Running parallel to this governance framework must be a sound enterprise data strategy. Only after both governance and data foundations are firmly entrenched should an organization pivot toward active experimentation and technological innovation.

Nandan strongly recommends carving out dedicated organizational capacity specifically for experimentation, enabling cross-functional teams to test emerging technologies and construct proofs of concept (POCs) without prematurely committing to enterprise-scale capital deployment. The pre-established governance frameworks and data strategies then provide objective criteria for determining which experimental initiatives warrant further financial backing and which should be immediately terminated. For Nandan, an internal innovation lab structured around these principles is the only way traditional enterprises can successfully keep pace with a technological landscape that evolves at lightning speed.

Crucially, this structured approach does not eliminate the inevitability of failure. Operational setbacks will occur, and an organization’s ultimate competitive differentiator is how rapidly its leadership can diagnose missteps and adapt accordingly. Reaffirming her focus on operational accountability, Nandan outlined a clear threshold for halting unproductive initiatives:

"If you see that none of the KPIs are moving, have the courage to stop and say, what do I need to change? We don’t need to keep pushing at something if things are not moving in the right direction. So a lot of it is strong leadership, along with all of the governance and the basic foundational capabilities in place."

When examining which institutions are successfully achieving real operating scale in sectors like BFSI and healthcare, Nandan notes that the industry remains in the nascent stages of enterprise AI maturity. Even the most advanced market leaders are currently experimenting, iterating, and frequently encountering failures. However, the organizations demonstrating the fastest progress share a common structural trait: highly concentrated AI decision-making authority.

Spreading artificial intelligence decision-making too broadly across a complex enterprise invariably backfires. Nandan cautions against decentralized operational models that generate organizational friction and conflicting mandates:

"When you say, I want democratization of AI, and I want everybody to work on it, and I want everybody to have opinions and everybody to make decisions, then I think that creates a lot of chaos and uncertainty, and then people end up having to report to five different managers. But when organizations create a tight AI unit and have strong leadership making very courageous and quick decisions without worrying about the optics, I think that is really helpful in moving things forward, so that we’re not wasting time on politics."

By centralizing responsibility for artificial intelligence within a dedicated, cohesive unit, enterprises can accelerate deployment velocity by eliminating competing decision paths and destructive internal politics. This focused operational model aligns seamlessly with Nandan’s governance-first philosophy: foundational guardrails are established by a centralized governing body possessing deep industry expertise and legal awareness, while executive leadership retains the clear authority required to execute decisive strategies within those protective boundaries. As financial institutions, insurers, and healthcare providers continue to deploy AI directly into the customer journey, this synthesis of bounded scope, unified data integrity, and centralized governance will ultimately define the boundary between sustainable digital transformation and operational failure.