The Inference
June 17, 2026
SEASON 1, EPISODE 2
In this episode of The Inference, Shri Nandan, VP of AI Products and Experiences at Comcast, joins host Michelle Hawley to discuss what it actually takes to build and scale AI inside one of the largest enterprises in the country.
Nandan shares how her team uses agentic frameworks, customer data platforms and cross-functional alignment to drive AI strategy at Comcast. The conversation covers the real costs of scaling AI beyond the pilot stage, how to prioritize ideas using a revenue-cost-KPI-resource framework and why governance has become the most complex part of the job. Nandan also explores where AI delivers the clearest value in customer experience and who bears accountability when AI agents go wrong.
Enterprise AI Needs a North Star
Enterprise AI Needs a Reason to Exist
According to Shri Nandan, VP of AI products and experiences at Comcast, enterprise AI strategy should not begin with the technology. It should begin with the customer problem.
Nandan leads a cross-functional team spanning engineering, product, customer experience and data platforms, and she says alignment depends on giving every team the same North Star: what the customer needs, what the business needs and what the technology can actually support.
Before moving an idea forward, Nandan asks why the company should do it, whether there is scientific rigor behind it, whether it can be repeated and whether it can scale to millions of customers. In practice, that means prioritizing use cases tied to measurable friction, like repeat customer contacts that drive churn and frustration.
Scaling AI Means Owning the Cost and Risk
The hard part of enterprise AI is what happens after the demo works. Nandan cautioned that AI costs more than many companies expect once cloud infrastructure, staffing, maintenance, governance, evaluations and operational support are included. Even an agentic system needs people to monitor, improve and maintain it.
That's why governance has become central to AI deployment, with teams needing clear guardrails around security, compliance, ethics, bias, customer data and vendor risk.
The same discipline applies to customer experience: AI can handle repetitive tasks, but emotional or complex moments should move quickly to humans. For Nandan, accountability ultimately sits with the teams that build these systems, backed by a central governing body that sets standards for engineering, evaluations, vendors and customer metrics.
Want the video version of this episode? Check out VKTR TV's The Inference.