

The AI Infrastructure Stack with Jennifer Li
159 snips Jul 11, 2025
Jennifer Li, General Partner at a16z, is passionate about the transformational power of AI infrastructure. She dives into the shifting landscape where AI reshapes software stacks, creating a demand for innovative middleware. The conversation touches on the critical components for AI workloads, the evolution of B2B SaaS tools, and the importance of comprehensive systems supporting AI applications. Jennifer also discusses the intricate relationship between AI agents and APIs, illustrating how improved technology enhances user experiences across industries.
AI Snips
Chapters
Transcript
Episode notes
AI Shaping Infrastructure Layers
- AI infrastructure is fundamentally shifting to support new workloads like low-latency image and video generation.
- Different AI modalities like language models and diffusion models require distinct infrastructure optimizations.
Layered Model Usage in Production
- Production AI stacks increasingly layer small deterministic models before invoking large language models.
- This layered approach optimizes costs and leverages each model's strengths effectively.
Automate Mundane Document Tasks
- Automate soul-crushing tasks like document transcription with AI to free humans for higher-level work.
- Use specialized smaller models to handle complex, domain-specific document parsing before passing data to larger models.