A MULTI-AGENT ORCHESTRATION ARCHITECTURE FOR RELIABLE TASK EXECUTION IN LARGE LANGUAGE MODEL SYSTEMS

Authors

Keywords:

Keywords: Agentic AI, Multi-Agent Systems, Large Language Models, Task Orchestration, LLM Reliability, Structured Outputs, Critic Agent, Workflow Automation.

Abstract

The rapid deployment of large language models (LLMs) into production environments has exposed a fundamental reliability gap: a single monolithic model prompted to plan, execute, and verify a complex task in one pass frequently produces plausible but incorrect results, with no internal mechanism to detect its own failures. This study proposes a multi-agent orchestration architecture in which a complex task is decomposed across specialized LLM agents — a Planner, one or more Executors, and a Critic — coordinated by a deterministic controller that manages state, routing, and termination. Each agent operates with a constrained role, a temperature of zero, and a structured output schema validated against a typed contract, allowing the controller to reason about agent outputs programmatically rather than through free-form text parsing.

The proposed controller treats task execution as a finite-state process: the Planner produces an explicit task graph, Executors resolve individual sub-tasks, and the Critic evaluates each result against acceptance criteria, triggering bounded retries or re-planning when validation fails. Experimental evaluation on a benchmark of 300 synthetic multi-step tasks shows that the orchestrated multi-agent system achieves a task-success rate of 0.91, compared with 0.68 for a single-agent baseline and 0.79 for a single agent augmented with chain-of-thought prompting. The orchestrated system also reduced unrecoverable failures by 71% through critic-triggered retries. Results indicate that explicit role separation, structured inter-agent contracts, and deterministic control flow provide a practical and auditable foundation for reliable LLM-based automation.

References

1.Jalolov, T. S. (2024). ИСКУССТВЕННЫЙ ИНТЕЛЛЕКТ С ИСПОЛЬЗОВАНИЕМ ФАЛЬШИВЫЙ ИНФОРМАЦИЯ ОПРЕДЕЛИТЬ МЕТОДЫ. Advanced methods of ensuring the quality of education: problems and solutions, 1(3), 53-59.

2.Sadriddinovich, J. T. (2024). BASICS OF PSYCHOLOGICAL SERVICE. PSIXOLOGIYA VA SOTSIOLOGIYA ILMIY JURNALI, 2(4), 61-67.

3.Jalolov, T. S. (2024). PYTHONNING MATEMATIK KUTUBXONALARINI O'RGANISH: KENG QAMROVLI QO'LLANMA. BIOLOGIYA VA KIMYO FANLARI ILMIY JURNALI, 2(5), 71-77.

4.Jalolov, T. S. (2024). РАЗВИТИЕ ТЕХНОЛОГИЙ ИСКУССТВЕННОГО ИНТЕЛЛЕКТА В САМОДВИЖАЩИХСЯ РОБОТАХ. Methods of applying innovative and digital technologies in the educational system, 1(2), 1-7.

5.Jalolov, T. S. (2024). ЭКОЛОГИЧЕСКИЙ СИСТЕМЫ ИСКУССТВЕННЫЙ В МОНИТОРИНГЕ ИНТЕЛЛЕКТ ТЕХНОЛОГИЙ ПРИЛОЖЕНИЕ. Advanced methods of ensuring the quality of education: problems and solutions, 1(3), 86-92.

6.Jalolov, T. S. (2024). MATNNI QAYTA ISHLASH ORQALI TIL O ‘RGATISH ILOVALARINI RIVOJLANTIRISH. Modern digital technologies in education: problems and prospects, 1(2), 103-108.

7.Jalolov, T. S. (2024). ЭФФЕКТИВНОЕ ИСПОЛЬЗОВАНИЕ ТЕХНОЛОГИЙ ИСКУССТВЕННОГО ИНТЕЛЛЕКТА В ЭКОНОМИЧЕСКОМ МОДЕЛИРОВАНИИ. Methods of applying innovative and digital technologies in the educational system, 1(2), 27-32.

8.Jalolov, T. S. (2024). AVTONOM ROBOTLARDA SUN'IY INTELLEKT TEXNOLOGIYALARINI RIVOJLANTIRISH. Ensuring the integration of science and education on the basis of innovative technologies., 1(3), 56-61.

9.Jalolov, T. S. (2024). СОЗДАНИЕ ДИАГНОСТИЧЕСКИХ СИСТЕМ НА ОСНОВЕ ИСКУССТВЕННОГО ИНТЕЛЛЕКТА В СИСТЕМЕ ЗДРАВООХРАНЕНИЯ. Methods of applying innovative and digital technologies in the educational system, 1(2), 14-20.

Jalolov, T. S. (2024). KIBERMUHOFAZANING TA'LIM JARAYONIDAGI O'RNI. PEDAGOGIK TADQIQOTLAR JURNALI, 2(1), 189-192

Downloads

Published

2026-06-09