In an era where new AI models, vendors, and frameworks seem to emerge weekly, organizations face a paradox. While access to advanced AI capabilities is becoming increasingly commoditized, the ability to turn those capabilities into durable competitive advantage remains elusive.

That tension sits at the heart of this conversation between Jyoti Chawla — a technology executive who has led large-scale modernization and enterprise transformation across global financial services and technology organizations — and Karen Odegaard, Partner and AI & Data Leader at Guidehouse.

With more than 25 years of experience driving complex digital and operating model transformations across Fortune 100 companies like IBM, Deutsche Bank, and Cisco, Chawla brings a pragmatic lens grounded in execution at scale.

Her core thesis is clear: “Sustainable AI advantage is not built on better models. It is built on better foundations.”

As enterprises move from AI experimentation to AI accountability, Chawla represents a new category of technology leadership — one that combines architectural depth, governance rigor, and strategic foresight brought to you as a part of the CDO Magazine interview series below.

Why data, not models, will decide who wins with AI

As Odegaard opens the discussion on data as a competitive advantage, Chawla quickly reframes the conversation around uncertainty. “The landscape is changing rapidly in terms of AI vendors, and models are being released every day,” she says. “We can’t project three years from now, which vendor and which model will be surviving.”

Drawing parallels to the early days of cloud computing, Chawla notes that while AI platforms will eventually rationalize around a smaller set of leaders, organizations cannot afford to wait for that clarity. “The smartest investments right now are fearlessly establishing good data infrastructure, sound fundamentals, and flexible architectures,” she explains.

In a world where foundational models are broadly accessible, Chawla argues that differentiation shifts elsewhere. “If everyone has access to the same foundational models, what becomes the differentiator?” she asks. “The secret sauce lies in the proprietary data and how effectively we can use it.”

Strong data management, governance, quality, and access, she adds, are what allow organizations to move beyond pilots and scale AI into industrial-grade applications.

Building the data flywheel through feedback and transparency

As organizations scale AI, maintaining data accuracy and relevance becomes a continuous process rather than a one-time fix. Chawla describes data quality as a living system. “I consider data quality as data having its own health,” she says. “It’s a journey.”

Reflecting on her time at Deutsche Bank, Chawla describes an approach centered on transparency as data moves across systems. “How about having a nutritional label for the data set?” she asks. “So the consuming systems understand what they’re consuming and the risks around it.”

That “label,” she explains, evolves alongside the data as quality improves, enabling downstream teams to make informed decisions. AI has dramatically accelerated this process by enabling metadata discovery, contextual enrichment, and anomaly detection at scale.

Beyond tooling, Chawla emphasizes operating principles that help organizations break silos. “Improve the quality at the source,” she says. “Bring DevOps principles into DataOps. Clean it up front, keep data where it is, and provide access where it needs to be.”

AI risk starts with non-determinism

As the conversation turns to risk, Chawla highlights why AI introduces fundamentally new challenges for enterprises. “We’ve deviated from deterministic, rule-based systems into non-deterministic systems,” she explains. “AI presents a black box.”

Bias, hallucinations, and unintended propagation of sensitive data are no longer theoretical risks. Addressing them requires more than traditional security controls. “It’s layering additional controls,” Chawla says, “especially as we look at agentic AI and agentic ops.”

She likens onboarding AI agents to onboarding new employees. “Someone said it’s like a teenager,” she notes. “You need human-in-the-loop onboarding. What rights do those agents have? Are they aligned with goals?”

Just as important is knowing how to intervene when things go wrong. “When an agent is not adhering to its goals,” she says, “how do you offboard it?”

Non-negotiables across the AI lifecycle

Asked about controls that enterprises cannot compromise on, Chawla points to principles that transcend regulation. “Privacy by design, minimizing the use of data to what’s required for the outcome, differential privacy, and encryption at all states,” she says.

Auditing and traceability are equally critical, especially as models are fine-tuned with proprietary data. “You don’t want to introduce new bias or model drift,” she explains. “Testing for bias is super important.”

While regulatory environments differ across regions, Chawla stresses that existing requirements like GDPR, data sovereignty, PCI, and HIPAA still apply. AI does not replace those obligations; it intensifies them.

Managing third-party and SaaS risk at scale

When AI solutions rely on external vendors, risk management extends well beyond procurement. “If you’re not compliant with data sovereignty,” Chawla says, “you probably lose your license to operate.”

She encourages organizations to evaluate the full stack, from operating systems and hyperscalers to backup locations and operational support models. “Where are backups stored? Where is support operating from?” she asks. “All of that matters.”

To manage complexity at scale, Chawla points to automation and policy-as-code approaches, such as Open Policy Agent, which can enforce regional controls and detect violations in real time. “That’s the only way to do it at scale,” she says. “Invest in that.”

Why hub-and-spoke operating models endure

On enterprise AI operating models, Chawla is clear about tradeoffs. Centralized teams offer efficiency but risk disconnecting from business outcomes. Fully federated models introduce coordination challenges. The model she has seen work best blends both.

“Platform investments should be centralized,” she explains, “but governance works beautifully as a hub-and-spoke.”

In this approach, a central team establishes shared standards and platforms, while domain teams act as ambassadors, translating those principles into business-specific use cases with clear ROI. “It creates a feedback loop,” Chawla says, “and makes you more nimble in adopting AI across the organization.”

Humans: The Greatest Catalyst for Responsible AI Adoption

While AI adoption is often framed as a technology challenge, Chawla sees people as the real inflection point. “No businesses are missing the opportunity to rethink using AI,” she says. “But adoption is where the work is – AI Adoption starts and scales with People”

At Cisco, as an executive sponsor for Agentic AI enablement communities, and communication, Chawla drove her approach on AI adoption to be centered on real-world usage. “Don’t just track training completion,” she says. “Look at adoption. Identify the champions, build the communities that amplify the learnings and communicate the real impact.” Hackathons and experimentation help spark interest, but long-term success requires discipline. “Don’t retrofit a use case for AI,” Chawla cautions. “Use AI as a tool for the use case.”

She also encourages leaders to be pragmatic about build-versus-buy decisions, especially as AI-first platforms mature. “If someone has already invested and done it well,” she says, “it’s fair to make that leap, as long as they’re in it for the long haul.”

CDO Magazine appreciates Jyoti Chawla for sharing her insights with our global community.