2025-04-10 [Case Study] Achieving Smart Factory Operational Excellence with AI Agents
The manufacturing industry is evolving at an unprecedented pace, demanding real-time intelligence, predictive capabilities, and adaptive decision-making. ASRock Industrial and Aimirim have partnered to implement IEC 61499-powered smart and open process for a global tobacco company, incorporating AI agent solutions that offer real-time suggestions and troubleshooting. By combing ASRock Industrial’s iEP-6010E edge AI device with Aimirim’s Shaman, this solution brings real-time AI agent to factory operations such as quality control, production, and maintenance, empowering operators with instant, intelligent recommendations to resolve production inefficiencies.
Challenges
British American Tobacco (BAT), a leading global tobacco company, faced significant challenges in process automation, preventing scalability and optimization in production. The factory’s process control relied heavily on human expertise, making it difficult to achieve consistent product production and inefficiencies.
Additionally, frequent machine stoppages and the lack of predictive maintenance resulted in higher operational costs, excessive raw material waste, and unnecessary energy consumption during reprocessing. Without a proactive approach to maintenance, the company struggled with unexpected downtime, negatively impacting production output. Furthermore, the factory’s reliance on a IEC 61131 based control system created limitations due to its centralized structure and lack of interoperability, hindering system expansion and the adoption of industry 4.0 and AI-driven technologies. As production needs grew, these outdated automation constraints became a major obstacle to efficiency and future scalability.
Solution
ASRock Industrial and Aimirim partnered to transform process automation at manufacturing facilities of the British American Tobacco, a global tobacco company. Through AI agents, they addressed key challenges in production consistency, operational efficiency, and automation scalability. By integrating ASRock Industrial’s iEP-6010E edge AI device, powered by NIVIDIA® Jetson Orin™ NX module, with Aimirim Shaman, the AI agent solution introduces real-time AI-driven process control, predictive maintenance, and an open, scalable automation framework that enhances quality, production, and maintenance operations.
To address the challenge of inconsistent product quality due to manual dependent process control, the AI agent structures machine sensor data, logs, and production events into AI-driven diagnostics. This eliminates reliance on operator expertise, standardizes process control, and ensures greater production stability. Using LLAMA trained LLM models, the system builds a relational structure between AI models and operational manufacturing scenarios, allowing for intelligent decision-making and process adaptation over time. To combat frequent production downtime and excessive resource consumption, iEP-6010E provides real-time AI inference at the edge, enabling low-latency troubleshooting and predicative maintenance. This minimizes unplanned downtime by detecting potential failures early and optimizing machine performance, which directly reduces raw material waste and energy consumption. Furthermore, by leveraging IEC 61499’s smart open process capabilities, the tobacco company has enhanced system scalability and seamlessly integrated AI-driven workflows, creating a resilient, future-ready automation system that optimizes production efficiency and long-term adaptability.
Benefits
The integration of ASRock Industrial’s iEP-6010E and Aimirim’s Shaman AI agent at
factory of the tobacco company has resulted in some key benefits:
- Enhance product quality and consistency: By automating process control through AI-driven diagnostics, the system has reduced process variability, ensuring a stable and repeatable production process. In the next phases, the goal is to achieve a reduction in process variability by 25%-56%.
- Optimized production efficiency and resource utilization: Predictive maintenance powered by AI agents has minimized unplanned downtime leading to a 5% reduction in raw material waste and an 4% decrease in fuel gas consumption, directly improving cost efficiency and sustainability. It means 31 Ton of CO2 reduction.
- More autonomy: The process was highly dependent on training, requiring over a year to prepare personnel to handle automation. With the new scenario, qualification now takes only 3 months of learning.
- Scalable and future-ready manufacturing: The adoption of IEC 61499’s distributed automation framework has provided greater flexibility, eliminating vendor lock-in and allowing seamless AI integration across industrial workflows, ensuring the tobacco company is positioned for continuous innovation and adaptability.
Hoberdan B. Silva is the digital operations manager for BAT in Uberlandia Factory, he specialized in automation, system integration and process optimization for the industry. According to him, the potential benefits in next phases are:
- Autonomous process (learning curve 1y⭢3m)
- Process Variability reduction (25%-56%)
- Waste Reduction (0.22%)
- ESG – energy reduction (3%-5%)
- Production UPDT reduction
- Storage optimal space usage (capex avoidance)
- Working Pattern Review
- Crew optimization
We are already in the future of Global Industry 4.0, proving results. To ensure competitiveness and sustainability, a strong automation and data collection foundation is key. With Open Process Automation, we can ensure a well-established health maintenance and operational process ("shit in - shit out") and define a measurement model in advance.
Related Product