Improving the Efficiency of Language Agent Teams with Adaptive Task Graphs

TL;DR

LATTE is a novel framework that enhances language agent team efficiency by dynamically constructing shared task graphs, reducing resource usage and errors compared to traditional fixed or unstructured methods.

Contribution

Introduces LATTE, a distributed systems-inspired framework enabling dynamic, shared task graphs for coordinating LLM teams with improved efficiency and adaptability.

Findings

LATTE reduces token consumption and wall-clock time.

LATTE decreases communication overhead and coordination failures.

LATTE matches or exceeds the accuracy of existing team coordination methods.

Abstract

Large language models (LLMs) are increasingly deployed in teams, yet existing coordination approaches often occupy two extremes. Highly structured methods rely on fixed roles, pipelines, or task decompositions assigned a priori. In contrast, fully unstructured teams enable adaptability and exploration but suffer from inefficiencies such as error propagation, inter-agent conflicts, and wasted resources (measured in time, tokens, or file operations). We introduce Language Agent Teams for Task Evolution (LATTE), a framework for coordinating LLM teams inspired by distributed systems, where processors must operate under partial observability and communication constraints. In LATTE, a team of agents collaboratively construct and maintain a shared, evolving coordination graph which encodes sub-task dependencies, individual agent assignment, and the current state of sub-task progress. This protocol maintains consistency while empowering agents to dynamically allocate work, adapt coordination, and discover new tasks. Across multiple collaborative tasks and a variety of base models, we demonstrate how LATTE reduces token usage, wall-clock time, communication, and coordination failures (e.g. file conflicts and redundant outputs) while matching or exceeding the accuracy of standard designs including MetaGPT, decentralized teams, top-down Leader-Worker hierarchies, and static decompositions.