Aegis: Taxonomy and Optimizations for Overcoming Agent-Environment Failures in LLM Agents

TL;DR

This paper introduces Aegis, a set of environment optimizations that significantly improve the success rates of LLM agents in complex real-world tasks without altering the agents themselves.

Contribution

It presents a taxonomy of agent-environment failure modes and proposes targeted environment optimizations to enhance agent success rates.

Findings

Environment optimizations improve success rates by 6.7-12.5%.

Analysis of 142 agent traces reveals 6 failure modes.

Aegis enhances environment observability, computation offloading, and speculative actions.

Abstract

Large Language Models (LLMs) agents augmented with domain tools promise to autonomously execute complex tasks requiring human-level intelligence, such as customer service and digital assistance. However, their practical deployment is often limited by their low success rates under complex real-world environments. To tackle this, prior research has primarily focused on improving the agents themselves, such as developing strong agentic LLMs, while overlooking the role of the system environment in which the agent operates. In this paper, we study a complementary direction: improving agent success rates by optimizing the system environment in which the agent operates. We collect 142 agent traces (3,656 turns of agent-environment interactions) across 5 state-of-the-art agentic benchmarks. By analyzing these agent failures, we propose a taxonomy for agent-environment interaction failures that includes 6 failure modes. Guided by these findings, we design Aegis, a set of targeted environment optimizations: 1) environment observability enhancement, 2) common computation offloading, and 3) speculative agentic actions. These techniques improve agent success rates on average by 6.7-12.5%, without any modifications to the agent and underlying LLM.