ALAS: A Stateful Multi-LLM Agent Framework for Disruption-Aware Planning
This solves the problem of scalable and resilient planning with LLMs for applications like job-shop scheduling, though it appears incremental as it builds on existing agent frameworks with targeted improvements.
The paper tackled the problem of LLMs struggling with transaction-style planning requiring ACID-like guarantees and real-time disruption recovery by introducing ALAS, a framework that addresses deficits like lack of self-verification and persistent state, resulting in new best results on job-shop scheduling benchmarks for static planning and strong performance in dynamic scenarios.
Large language models (LLMs) excel at rapid generation of text and multimodal content, yet they falter on transaction-style planning that demands ACID-like guarantees and real-time disruption recovery. We present Adaptive LLM Agent System (ALAS), a framework that tackles four fundamental LLM deficits: (i) absence of self-verification, (ii) context erosion, (iii) next-token myopia, and (iv) lack of persistent state. ALAS decomposes each plan into role-specialized agents, equips them with automatic state tracking, and coordinates them through a lightweight protocol. When disruptions arise, agents apply history-aware local compensation, avoiding costly global replanning and containing cascade effects. On real-world, large-scale job-shop scheduling benchmarks, ALAS sets new best results for static sequential planning and excels in dynamic reactive scenarios with unexpected disruptions. These gains show that principled modularization plus targeted compensation can unlock scalable and resilient planning with LLMs.