Intelligence Squared What Are The Essentials for Reimagining Work with AI Agents?
30 snips
Oct 29, 2025 
In this discussion, Aaron Levie, co-founder and CEO of Box, shares insights on the future of work with AI agents. He emphasizes that organizations are facing a data problem rather than an AI problem. Levie explains how AI agents can unlock value from unstructured data and highlights the importance of governance and security in leveraging these tools. He encourages lean experimentation and scalability, while also addressing the need for AI literacy among teams. Ultimately, he envisions a future where AI reallocates tasks, freeing humans for more creative work.
AI Snips
Chapters
Transcript
Episode notes
Agents Do Multi-Step Workflows
- AI agents loop through tasks repeatedly rather than respond to single prompts, enabling them to complete multi-step workflows.
- This shift lets agents read many documents, store temporary memory, and produce long-running outputs that humans couldn't feasibly do alone.
Unstructured Data Is Untapped Gold
- About 90% of enterprise data is unstructured: contracts, notes, images, videos and more.
- Large language models can extract structure and surface insights from this previously untapped goldmine.
Make Data Strategy First
- Start your AI strategy at the data layer: define clear sources of truth and single systems of record per data type.
- Simplify access controls and governance before deploying many agents to avoid wrong answers and security leaks.