Agentic AI is moving beyond assistive tools toward systems that can reason, plan, and act within complex workflows. In the latest episode of the Harvard Data Science Review Podcast, we speak with Dirk Hofmann and Ulla Kruhse-Lehtonen, co-founders and co-CEOs of DAIN Studios, about what this shift means for organizations in practice.

The conversation explores how agentic AI differs from traditional automation, why outcomes matter more than outputs, and how humans and AI agents can work together responsibly. Drawing on their long-standing work in data and AI strategy, Hofmann and Kruhse-Lehtonen offer practical insights into strategy, governance, and the evolving “mindware” required to make agentic AI deliver real value.

The episode also highlights their forthcoming HDSR article, “The Agent-Centric Enterprise: Why 2–10x Productivity Gains Demand Radical Workflow Redesign,” and their joint online course with the Harvard Data Science Initiative, Agentic AI: Contextualized and Applied, which focuses on applying agentic systems responsibly in real organizational settings.

Our guests:

• Dirk Hofman is the co-founder DAIN Studios and CEO of DAIN Studios Germany
• Ulla Kruhse-Lehtonen is the co-founder of DAIN Studies and CEO of DAIN Studies Finland

In this month’s episode of the Harvard Data Science Review Podcast, we explore the rapidly evolving concept of digital twins—dynamic, data-driven replicas of complex systems—and their growing influence across engineering, cities, healthcare, and society at large. Blending real-world case studies with big-picture insight, the discussion highlights how real-time data, sophisticated models, and massive computing power converge to let us safely test ideas, anticipate disruptions, and design smarter systems. Just as importantly, the episode tackles the critical questions of ethics, privacy, and public trust, making it an essential listen for anyone interested in where data science is headed—and how it can responsibly shape the world we live in.

Our guests:

• Rachel Franklin is the executive director of the Center for Geographic Analysis at Harvard University
• Patrick Johnson is the executive vice president of Corporate Research and Science at Dassault Systèmes

Fine wine meets data science in this month’s episode of the Harvard Data Science Review Podcast. Hosts Liberty Vittert Capito and Xiao-Li Meng explore how data, taste, and technology intertwine with Eric LeVine, founder of CellarTracker, the world’s largest community-driven wine database. What began as a personal project for managing a home cellar has evolved into a global platform with millions of users and billions of data points on wines, prices, and human preferences.

During the conversation, they unpack the origins of CellarTracker at Microsoft, how big data and machine learning reveal trends in taste and behavior, and the use of AI to predict “Will I like this wine?” They investigate how to find your digital wine twin, data quality and privacy, and how AI could change how we buy, drink, and enjoy wine without losing the human touch.

Grab a glass and join us for a thoughtful conversation about curiosity, community, and enjoying life—responsibly.

Our guest:

Eric LeVine is president and CEO of CellarTracker, the world’s most comprehensive database. Previously he was a group program manager at Microsoft.