Everything on AI/ML Research

Reinforcement learning on agentic-native search tools outperforms training on traditional SERP wrappers in efficiency and accuracy.

Impact: Developers should prioritize agentic search APIs for RL workflows to maximize agent performance and reduce token consumption.

— from Exa CEO on Agentic Search, Tokenpocalypse, and Beating Google · a16z Podcast· Jun 06, 2026