Scaling Agentic Inference Across Heterogeneous Compute [Zain Asgar] - 757
The TWIML AI Podcast with Sam Charrington · 47:44 · Watch on YouTube
Gimlet Labs CEO Zain Asgar argues that AI inference, especially agentic workloads, gets much cheaper and faster when it's split across mixed hardware (different GPU generations and vendors like Nvidia, AMD and Intel,...