The Power of Digital Policy

Kevin Surace joins Kristina to talk about what changes once AI stops giving advice and starts acting. They cover the difference between permissions and authority, why so many companies have agents running that IT doesn't even know about, and why biometric approval is becoming the only real way to prove a human signed off on something.

Companies are pouring money into AI to make work faster and more predictable, but the world keeps producing uncertainty anyway. Mary Lemmer, who works at the intersection of improv, leadership, and organizational behavior, joins Kristina to talk about why the systems built to reduce risk often end up punishing the very experimentation that innovation depends on. She explains how applied improv trains people to stay functional when there's no playbook, what "yes, and" actually means when your boss hands you a project you disagree with, and why a leader's ability to regulate their own reaction matters more than having the answer.

What happens when AI stops recommending and starts spending? Rodrigo Coelho, CEO of Edge & Node and a founding member of Ampersand, joins Kristina to talk about agentic payments, where the money is gone the second it moves and the policy which must envelope the transaction. The conversation delves into why permission isn't the same as authorization, why policy checks belong at the wallet level rather than inside the LLM, and who's actually accountable when an agent gets it wrong. If you're thinking about identity, authorization, observability, and governance for AI agents, this one's for you.

Thomas Hill, Senior Director of Digital and Commercial AI Enablement at Danaher, joins Kristina Podnar to share what it takes to build an enterprise commercial AI program from scratch across 15+ operating companies. From unifying eight million contacts and three million customers into a governed golden record, to navigating GDPR and cross-company data sharing, to embedding policy directly into the agentic orchestration layer, Thomas explains why the hardest part of AI isn't the AI — it's the data readiness, stakeholder alignment, and work redesign that make it trustworthy and scalable.

AI is forcing organizations to confront a reality they have often ignored for years: information quality matters. In this episode, Kristina Podnar sits down with taxonomist and knowledge management expert Ian Davis to explore why concepts like taxonomy, ontology, semantic modeling, and knowledge graphs have become critical foundations for successful AI deployment. Together, they discuss how AI exposes weaknesses in fragmented information environments, why metadata and information architecture are emerging as strategic business assets, and what leaders must do to create the governance, stewardship, and organizational discipline needed to scale AI effectively. The conversation also examines the growing importance of human oversight, the role of semantic structures as AI guardrails, and why the long-overlooked professionals responsible for organizing enterprise knowledge may be among the most important contributors to successful AI adoption.

AI isn’t slowing down because of weak models or poor prompts—it’s being held back by something far less flashy: data privacy. In this episode, Kristina sits down with Adam Kamor of Tonic.ai to unpack why organizations are struggling to safely use their own data, how unstructured data creates massive operational friction, and what it actually takes to move from manual redaction to scalable, privacy-first AI systems. The conversation challenges the idea that privacy is just a compliance checkbox—and reframes it as the key to unlocking AI velocity.

Kristina Podnar talks with data and analytics leader Ioana Mazare about why synthetic data is gaining traction, where organizations risk getting it wrong, and why strong governance is still essential. They explore privacy, bias, fit-for-purpose use cases, third-party synthetic data providers, and the leadership, strategy, and accountability needed to make data trustworthy in the age of AI.

AI is no longer just generating content—it’s retrieving, reasoning, and increasingly making decisions. And yet, many organizations are approaching it as if better prompts or better models will solve everything. In reality, AI is only as effective as the content it relies on. The more autonomous these systems become, the more they depend on content that is structured, connected, and governed with intent.

In this conversation, Kristina Podnar sits down with Carrie Hane to unpack what that actually means in practice. They explore the difference between formatting and true semantic structure, why duplication and inconsistency quietly undermine AI performance, and how weak or nonexistent content models introduce risk at scale. They also get into a question that doesn’t get asked often enough: who actually owns the content model in an organization—and what happens when no one does?

This episode is a reminder that AI doesn’t eliminate the need for structure—it exposes where it never existed. If organizations want AI to perform reliably, content can no longer be treated as a byproduct of publishing. It has to be treated as infrastructure.

Kristina Podnar reconnects with Brian Bauer to explore what’s changed in just a few months as agentic AI moves from experimentation to execution in banking and other regulated industries. They unpack the “year of reckoning” for AI ROI, why compliance can accelerate innovation instead of blocking it, and how organizations can use AI to reduce costs, increase revenue, and compete at the speed of business—without sacrificing prudence.

In this episode, Kristina Podnar sits down with Camille Stewart Gloucester to unpack what’s durable in today’s AI policy landscape — and what’s just noise. Tune is for why AI governance isn’t about regulation alone, but about infrastructure: decision rights, data access, organizational design, and continuous learning. Camille shares what she’s seeing across industry and government, where organizations are overconfident, and why retrofitting governance gets exponentially harder once AI systems are embedded. If you’re trying to balance speed, risk, and performance without defaulting to paralysis, this conversation is for you.