Derick Jose, MD - Industrial AI, Accenture, in an interaction with Asia Manufacturing Review, provided insight into how Industrial AI is impacting operations across industries. Derick shared his views on how predictive maintenance is evolving from just reducing downtime to optimizing spare parts inventory and workforce scheduling. He described how AI could create impact in asset reliability, product quality, health and safety, and environmental outcomes. Derick pointed out that mid-size manufacturers can start their AI journey in an affordable way through existing assets such as video surveillance, showing that Industrial AI is not just about profitability and performance, but fundamentally about creating sustainable and resilient operations.
With extensive cross-industry expertise in Energy, Manufacturing, and Engineering, Derick Jose is the Co-founder of Flutura and currently serves as Managing Director – Industrial AI at Accenture. Derick has driven the adoption of AI-enabled solutions into industrial operations and measures business outcomes for global leaders - Shell, TechnipFMC, Lupin, and ABB. As Managing Director, he has deployed Flutura's flagship Industrial AI platform - Cerebra - in over 20 countries and has driven over $300 million in impact and earning recognition from Gartner Peer Insights as a top-ranked solution provider.
How can industrial AI improve predictive maintenance beyond reducing downtime? Can it also optimize spare parts inventory and workforce scheduling?
Industrial AI has the potential to transform the way organizations manage and optimize their operations, delivering measurable impact across three key dimensions: asset reliability, product quality, and health, safety, and environmental outcomes. The first and perhaps most established domain is maintenance and asset reliability. By using advanced analytics, AI can detect equipment failures before they occur, which will ultimately improve the Mean Time between Failure (MTBF) and reduce the Mean Time to Repair (MTTR). The predictive capability increases uptime and productivity, while optimizing spare parts inventory to ensure organizations have the components they need exactly when they need them, with no excess capital being used in stock. As a result, the leaner maintenance functions that reduces costs and increases efficiency.
Equally important is the use of AI to manage quality in core operations. In discrete or continuous manufacturing, it is essential to have a high-quality output. A rejected part does not just mean waste; rejected parts can impact profit margins, reduce market-share, and negatively affect stock performance. AI enables manufacturers to dramatically reduce defects, protect consumer confidence, and maximize profit by able to detect and predict quality deviations in real-time and enabling proactive intervention.
The third area where industrial AI can provide value is health, safety, and environment compliance. In regulated industries, exceeding emission limits of methane or carbon dioxide, leading to substantial fines, reputational risk, and legal consequences. Using AI-based monitoring and control systems, organizations can be ahead of regulatory requirements and detect anomalies early in the process and take action to minimize emissions as well as ensure health and safety in the work environment.
Industrial AI increases revenue and profitability and it directly enhances operational resilience with more reliable assets, improved quality, and healthy, safe and environmentally reputable operations. This resilience is critically important to a sustainable industrial transformation.
How does industrial AI handle the challenge of integrating data from legacy equipment that may lack modern sensors?
To develop an industrial data lake requires the integration of multiple diverse streams and sources of information originating from many operational systems. The operational data lake is built from ingestion of data from SCADA systems, MES systems, and process historians that represent the fundamental activities and functions of the industrial and manufacturing processes. Getting data directly from the equipment to a data lake can be challenging. Here industries can use data adapters as a bridge. These data adapters ensure seamless flow of data from machines, sensors, and control systems, into a central data repository.
However, time-series data is not enough to create a holistic view. To gain more meaningful insights, time-series data must be combined with the maintenance record, which provide a historical perspective of the equipment’s performance, failures and service interventions. By aligning operational signals to maintenance history, organizations can move from reactive fixes to predictive and prescriptive maintenance strategies.
Additionally, lab data from instrumentation systems must be incorporated. Quality testing, compliance measurements, chemical analysis are typically follow their own sets of data silos yet their integration with process data enriches the overall operational intelligence. Although structured datasets remain useful, organizations are more aware of unstructured sources such as audio and video. Inspection video feeds, or monitoring devices paired with process data offer a visual way of identifying anomalies. Similarly, the voice of technicians conveying experiential knowledge and real-time observations can be digitized and analyzed to provide valuable context. Many inputs as a group offer a 360 degree view of the operation. It can be a valuable tool for decision makers in improving efficiency, increasing safety, decreasing downtime, and supporting commercial innovation.
How can mid-size manufacturers with limited data maturity start their AI journey without large upfront investment?
For small or medium-sized manufacturers, the journey toward digital transformation does not have to begin with significant investments in sensors, automation systems, or industrial machinery. The simplest beginning can be leveraging resources that already exist on the shop floor. Most manufacturing plants typically have video surveillance cameras for security and general surveillance cameras for basic security and monitoring purposes. These cameras, when repurposed through video analytics, can maximize operational efficiency.
It is simple to repurpose existing resources. Manufacturers already have video surveillance on their shop floor, the action of installing new sensors or counting devices to identify actions or activities can be as simple as mining existing video surveillance videos. With the implementation of computer vision algorithms, it is possible to identify the time taken for each task, observe workflow patterns, and uncover operational inefficiencies without any act of a sensing device. For instance, a system would be able to identify how long it takes a person to move material between stations or how often a bottleneck exists for a step in their operations.
Such analytics can highlight opportunities for improvement without disrupting operations. The history of the data shows patterns that will guide managers to redesign workflows, minimize idle time, and maximize labor and machine usage. Video analytics are also capable of contributing to safety and compliance monitoring. From the analysis of motion, posture, and even compliance with safety protocols, organizations can minimize the potential for down time, while creating a safer workplace.
All of this can be done without significant upfront investment, as many small manufacturers are concerned about the cost to digitally transform. In summary, capitalizing on existing video systems can allow manufacturer to take their first steps into Industry 4.0 without a lot of capital investment. Thus, it illustrates how even limited resources combined with intelligent algorithms can produce results for increasing efficiencies and productivity.