DeepFactory: Multi-Agent System for Industrial Production Management and Optimization

The manufacturing industry faces efficiency and integration challenges, even with advancements in Industry 4.0, IoT, and data analytics. Despite the availability of tools for different stages of the production process, management remains fragmented, resulting in uncoordinated decisions, bottlenecks, and low adaptability. Therefore, this work proposes DeepFactory, a multi-agent platform that integrates material planning, sequencing, execution, maintenance and quality control. The platform combines optimization, machine learning, and computer vision models, including inventory coverage, sequencing via mixed integer programming, predictive maintenance, and automated inspection, coordinated by AI agents. As a differential, the structured integration of these paradigms in a stage-oriented architecture, with the generation of actionable insights, stands out. As a system report, this work presents an illustrative use case in a simulated scenario of pharmaceutical manufacturing, demonstrating the applicability of the proposed architecture.