Focus on Second Open Call winners: Optimizing fermentation and boiling processes in a microbrewery through edge AI analytics– EdgeAI4Beer

WHAT ABOUT THE EXPERIMENT?

The EdgeAI4Beer experiment is a real-world demonstration of how artificial intelligence and edge computing can transform production processes in the traditional craft beer industry. Implemented at Paragon Microbrewery, a small-scale, regionally rooted brewery in Greece, the project integrates AI capabilities directly into the brewery’s daily operations—specifically targeting the fermentation and boiling stages of the brewing process. The motivation behind the project arises from recurring challenges faced by small brewers: high dependency on manual monitoring, inconsistency in batch quality, limited automation, and lack of real-time predictive insights. These limitations can lead to inefficiencies, product variability, and missed opportunities for optimization. EdgeAI4Beer addresses these “pains” by embedding AI intelligence close to the production floor, enabling fast, data-driven decisions without requiring costly cloud infrastructure or complex IT integration.

The core concept involves the development and deployment of two AI models:

1. A fermentation forecasting model, using a Random Forest Regressor enhanced with conformal prediction, deployed on a Jetson Orin Nano edge device, predicts when a batch will complete fermentation and provides confidence intervals to assist scheduling.

2. A boiling anomaly detection model, based on an autoencoder neural network, is embedded in a PLC-powered Arduino, detecting abnormal behavior in real time and issuing alerts.

By distributing intelligence at the edge, the system ensures low latency, resilience to connectivity issues, and privacy of operational data key benefits for SMEs operating outside large industrial hubs. The architecture includes local dashboards, sensor integration (e.g., Plaato Pro) and lightweight APIs.

The site of the experiment allows EdgeAI4Beer to explore the full potential of edge AI in a resource-constrained yet highly specialized environment. One of the central challenges tackled by the project is achieving seamless integration between humans and machines. The system is built to assist human operators by providing accurate, interpretable insights that enhance their decision-making. This approach encourages trust in AI, facilitates gradual digital adoption, and maintains the craftsmanship central to the brand’s identity.

WHAT IS THE EXPECTED IMPACT?

The EdgeAI4Beer experiment is expected to deliver significant impact across multiple business and technological dimensions within the craft brewing sector. By integrating AI and edge computing into the fermentation and boiling processes, the project will enhance production predictability, product quality, and operational efficiency.

On a business level, the AI-enabled fermentation forecast will reduce batch variability and improve scheduling accuracy, directly increasing tank utilization and product consistency. The anomaly detection system in the boiling stage will allow real-time identification of process deviations, leading to reduced downtime, fewer defective batches, and lower maintenance costs. The project will reduce manual interventions and repetitive tasks, freeing up skilled staff to focus on high-value activities such as recipe innovation. This will improve employee satisfaction and enhance the company’s image as a digitally mature, forward-looking craft producer. Additionally, real-time dashboards and AI outputs will empower operators with greater process control and decision confidence.

From a technological standpoint, the experiment contributes to the democratization of AI for SMEs. By demonstrating the feasibility of low-cost, scalable, and standards-compliant edge AI solutions, EdgeAI4Beer will serve as a replicable model for other small manufacturers across Europe.

The open and modular architecture will also feed into the AI-on-Demand (AIoD) ecosystem, promoting reuse and knowledge sharing. Ultimately, the project supports the AI REDGIO 5.0 mission by enabling human-centric, interoperable, and regionally distributed AI deployment in traditional industries.