Featured research · 2025
AI Agents Need Managers, Not Just Prompts.
A research presentation backed by 60+ academic papers, introducing the Compensation Engineering framework.
Most failures in AI agent deployments aren't model failures — they're management failures. Agents are tasked without clear acceptance criteria, run without independent verification, and corrected without feedback loops. This research catalogues twelve documented failure patterns from the academic literature and maps each to an evidence-based management practice borrowed from human team management.
