AI in Action The power of AI starts with better data
11 snips
Nov 18, 2025 Madison Faulkner, a venture partner and former data/AI operator with experience at Meta, discusses the transformative power of data for AI. She emphasizes that data chaos can be turned into clarity through effective governance and strategy. Madison dives into the importance of structuring unstructured data with knowledge graphs, the pitfalls of misaligned data strategies, and the emerging trend of treating data as code. She also shares insights into evaluating data startups and predicts a future of more autonomous AI systems driven by robust data foundations.
AI Snips
Chapters
Transcript
Episode notes
From Banking To Moonshot Data Work
- Madison Faulkner left a planned banking career to join early data work at Meta and helped shape nascent data science roles.
- She moved from analytics to deep learning infrastructure and worked with Facebook AI Research on moonshot projects.
Data Is The Real Unlock For Agentic AI
- Data is the unlocking asset for agentic AI and enterprise value creation, yet it lacks focused innovation attention.
- Unstructured data only creates value when tied to a structured semantic layer like a knowledge graph.
Build A Business-Centered Knowledge Graph
- Build a knowledge graph that models your business entities and connects products to users for contextualized retrieval.
- Maintain product and user embeddings to enable multidimensional retrieval and dynamic recommendations.