Inference accelerated small models as tools for big agents

Lightning talk (5:55–6:05 pm) on why frontier models don't make sense for tasks already saturated at 99%+ accuracy; hosted by Daytona & AWS Builder Loft, San Francisco.