The Joe Reis Show The Agentic Future: How Streaming is Evolving for AI w/ Tyler Akidau
Oct 15, 2025
In this engaging conversation, Tyler Akidau, a seasoned streaming systems engineer and CTO at Red Panda Data, unveils the transformative impact of AI on data infrastructures. Tyler shares his journey from AI skepticism to excitement, discussing how agentic AI systems will reshape business operations. They dive into the reimagining of data architectures for machine-speed AI, the need for rigorous auditing, and the shift from windowing to a table-based approach. Tyler also explores governance challenges and the chaotic nature of AI agents using a clever D&D analogy.
AI Snips
Chapters
Transcript
Episode notes
Tyler's Streaming Origins At Google
- Tyler started on MillWheel at Google and later helped bring streaming semantics into Cloud Dataflow and Beam.
- He wrote popular O'Reilly posts and a book that helped spread streaming ideas.
Streaming Evolved Toward Consistency
- Streaming matured from speed-first systems to strongly consistent, production-ready platforms.
- The hardest part is making streaming reliable and cost-effective under the hood.
Tables Over Windowing For Usability
- Declarative table-like APIs let users ask for outputs and let the system handle incremental computation.
- This approach broadens streaming to business analysts via SQL-style interfaces.