Yogesh Kawadkar

Giới thiệu

Academic / Researcher at Daimler Truck. 3–5 years of experience with AI and ML. Interested in consumer AI: products, apps and end-user experiences. Registered as a builder for Agentic AI Build Week.

1. Three things, all pointing at the same thesis — agents get useful when they have memory and permission, not when the…

   Three things, all pointing at the same thesis — agents get useful when they have memory and permission, not when the model gets bigger. Main one: a vertical agent for procurement ops, handling the messy email + spreadsheet back-and-forth between buyers, vendors, and finance. One design partner pilot, turnaround time on vendor threads down from ~2 days to a few hours. On the side: an AI-native outbound experiment (reply rate 7% → 22%, the real unlock was timing not copy), and a long-horizon memory layer for coding agents that persists "what was tried, what worked, what failed" across sessions…

Kỹ năng và ngành