Data Automation Is Driving Discrete Manufacturers toward Enhanced Prescriptive Analytics and Generative AI Deployments

By James Iversen | 4Q 2024 | IN-7556

Registered users can unlock up to five pieces of premium content each month.

Log in or register to unlock this Insight.

Data automation is the process in which large datasets are Extracted, Transformed, and Loaded (ETL) into a cloud-based repository for further analysis without human intervention. This process was historically time-intensive and error-prone due to the manual upload of machine data by data scientists. However, as more machines, Internet of Things (IoT) devices, Programmable Logic Controllers (PLCs), and software are being added to the factory floor, unstructured datasets are becoming larger and untenable for humans to manage with the total worldwide data generation in manufacturing sitting at 2.1 Zettabytes (ZB) in 2024, reaching 4.4 ZB by 2030.

Data automation is becoming especially prevalent in discrete industries, which produce 61% of all manufacturing data, half of which comes from the automotive sector (see ABI Research's Data Generation in Industrial Environments Forecast Market Data Overview: 3Q 2024 (PT-3149)). Companies such as Alteryx, Databricks, HighByte, IBM, Red Hat, and Snowflake are reacting to increased demand for data scientists manually uploading machine data by developing solutions that standardize data pipelines for better contextualization.

Data automation provides value to discrete manufacturers via two avenues: productivity gains through time-saving, and enhanced data quality. Data automation solutions reduce the overall time spent organizing and processing data by data scientists, allowing for more time spent on value-added tasks such as utilizing data for predictive analytics. In terms of data quality improvement, the removal of manual inputs such as coding ETL pipelines reduces the risk of human error so accuracy and integrity can be maintained throughout the process of manufacturing large-scale assemblies.

Discrete manufacturers also benefit from data automation through the deployment of Generative Artificial Intelligence (Gen AI). Gen AI models, which have most recently been deployed on the factory floor as large language copilots, must be trained on accurate machine data with access to real-time analytics. To gain insight and preventative actions into machine health and operational status through natural language inputs, up-to-date datasets must be used as training sources to ensure Gen AI models provide accurate responses. With data automation, precise models can be built out with reliability and assurance that responses will be backed by either factory floor data or Standard Operating Procedure (SOP) manuals.

Data automation is also more adaptable than traditional manual uploads and can leverage inputs outside of standard text such as images and videos, which bolster Gen AI models. Data automation is playing a pivotal role in providing the necessary data to train Gen AI models and the importance of having a strong data foundation will only increase as Gen AI realizes its potential to produce US$12.6 billion in efficiencies in manufacturing by 2034 (see ABI Research's Generative AI Solutions for Discrete Manufacturing presentation (PT-3233)).

While both discrete and process manufacturers have utilized data analytics solutions for decades to track metrics such as Overall Equipment Effectiveness (OEE) and Total Effective Equipment Performance (TEEP), the manufacturing industry is looking past descriptive analytics and toward prescriptive use cases. Today, 67% of manufacturers have the capability to collect and analyze data in near-real time, yet only 33% can perform prescriptive analytics, which is the end goal for manufacturers as profit increasing operations such as root-cause analysis and predictive maintenance can provide a tangible Return on Investment (ROI) for data analytics solutions.

This trend of automating data capture and upload for prescriptive analytics is further backed by ABI Research's recent survey of 461 manufacturing decision makers across the United States, Germany, and Malaysia, who ranked the removal of paper and digitalizing data collection as the second most important investment area over the next 12 months, beaten out by Robotic Process Automation (RPA).

Data automation is a crucial step manufacturers need to take before making the leap from descriptive to prescriptive analytics. The foundation for prescriptive analytics requires two core capabilities, the standardization of large datasets, and the real-time upload of contextualized machine data. Both prerequisites are provided by data automation and have the capacity to scale to all sizes of factories without increased manpower. With the deployment of automated data analytics solutions, manufacturers are able to perform vital functions such as predictive maintenance, root-cause analysis, inventory management, and sustainability checks.

One manufacturer that has been making tremendous strides in improving operations through data automation and the use of predictive analytics is Volvo with its use of Databricks Workflows. Using data automation through Workflows, Volvo solved its biggest challenge of slow upload speeds due to manual data entries, which allowed data to become obsolete and irrelevant before it could be utilized for analytics. Volvo has been scaling the deployment of Databricks Workflows across its entire operation for improved insight into warehouse allocations, spare part identification, and real-time machine report monitoring, which has resulted in an increase of 40% in efficiency gains and the availability of data scientists to focus on more value-add operations.