@KyleHovance @typesafeai Real-time agents: LLM sets high-level goals while Jev e

@KyleHovance @typesafeai Real-time agents: LLM sets high-level goals while Jev evaluates state every ~100ms to pick typed actions (move, shoot, dodge) without text gen delays. Sup

@KyleHovance @typesafeai Real-time agents: LLM sets high-level goals while Jev evaluates state every ~100ms to pick typed actions (move, shoot, dodge) without text gen delays. Support workflows: Jev classifies intent + routes/escalates/refunds with calibrated probs in a tight loop; LLM only drafts