AI Agents and RL Environments

Scale AI CEO on Meta’s $14B deal, scaling Uber Eats to $80B, & what frontier labs are building next

Topic 4 of 7

Published Oct 9, 2025 · Clip 17:03–23:34 (6:31)

Jason discusses RL environments as sandboxes for AI agents, emphasizing generalizable data across permutations. Examples include Salesforce navigation and calendar tasks. He stresses training for reliability and human handoff in low-accuracy scenarios.