The Data Stack Show 217: Bridging Data Models with Business Intuition with Zenlytic’s Founders Ryan Janssen and Paul Blankley
7 snips
Nov 27, 2024 In this engaging conversation, Ryan Janssen and Paul Blankley, co-founders of Zenlytic, share their journey from academia to building AI-powered business intelligence tools. They discuss the intersection of AI and data, the evolution of language models, and the challenges in model training. The duo emphasizes the importance of balancing data with intuition in decision-making and highlights the significance of multimodal data insights. They also touch on the fun behind naming their AI agent and the future of AI in analytics.
AI Snips
Chapters
Transcript
Episode notes
Early Language Models
- Paul and Ryan studied language models at Harvard, witnessing the rise of transformers.
- Early models were basic, but the progression to GPT-3 showed significant improvement.
Scaling Language Models
- Language models improved with increased compute and data, but those scaling opportunities are now limited.
- The next step is expandable inference time, leading to the emergence of AI agents.
Founding Zenlytic
- Ryan and Paul founded Zenlytic after seeing data often go unused after setup.
- Their goal is to solve the "last mile" problem and make data more accessible.