Beyond Autonomy: A Dynamic Tiered AgentRunner Framework for Governable and Resilient Enterprise AI Execution
For enterprise AI deployment, this work addresses the critical need for governable agent execution, but the contribution is incremental as it applies known software engineering principles (separation of powers, tiered resource allocation) to the agent domain.
The paper addresses the lack of governability in LLM agent frameworks for enterprise deployment by proposing a Dynamic Tiered AgentRunner with risk-adaptive tiering, separation of powers, and resilience-by-design. The framework achieves Pareto-optimal safety-efficiency trade-offs, though no concrete performance numbers are provided.
Current large language model agent frameworks prioritize autonomy but lack the governability mechanisms required for enterprise deployment. High-risk write operations proceed without independent review, complex tasks lack acceptance verification, and computational resources are allocated uniformly regardless of risk level. We propose the Dynamic Tiered AgentRunner, a controlled execution protocol distilled from a production-grade multi-tenant SaaS platform. The framework introduces three core mechanisms: (1) Risk-Adaptive Tiering that dynamically allocates computational resources and review intensity based on task risk profiles, achieving Pareto-optimal trade-offs between safety and efficiency; (2) Separation of Powers architecture where proposal, review, execution, and verification are performed by independent agents with physically isolated boundaries; and (3) Resilience-by-Design through a Verifier-Recovery closed loop that treats failure as a first-class system state. We formalize the tier selectio