At INFOCUS 2026, Massimiliano Bartolozzi, Chief Information Officer at Italian tissue paper manufacturer Lucart, demonstrated how organizations can combine generative AI with existing JD Edwards EnterpriseOne tools to create powerful operational dashboards without investing in additional software.

His presentation, with contributions from Lucart’s Michele De Giacomo, “AI-generated dashboard for JDE,” explored how Lucart transformed its approach to batch job monitoring, resource optimization, and operational analytics using Claude AI, JD Edwards Orchestrator, and the native EnterpriseOne Classic Page. The project delivered measurable improvements within its first week while establishing a foundation for more advanced AI-driven business processes.

The Challenge: Managing 33,000 Daily Batch Jobs

Lucart operates a complex manufacturing environment spanning 12 plants, five lines of business, and customers in 70 countries. Its factories and logistics centers operate around the clock, making system availability essential. Despite different business processes across its operations, the company maintains a single JD Edwards instance and database.

One significant challenge involved managing approximately 33,000 batch jobs every day. The IT team needed to determine whether those jobs were necessary, how efficiently they ran, and whether their execution schedules created unnecessary infrastructure demands.

Previously, analyzing batch activity required manually extracting information from JD Edwards, exporting it into Excel, and building pivot tables or cross-reference analyses. This process was time-consuming and depended on knowing which questions to ask before beginning the investigation.

Bartolozzi wanted a more efficient approach that could immediately identify underutilized resources, unnecessary jobs, execution bottlenecks, and opportunities to redistribute workloads.

The requirement was straightforward: Build a modern analytics solution using tools already available within JD Edwards, without purchasing another software platform.

Building AI-Generated Dashboards with Existing JDE Tools

Lucart’s solution combined three technologies: Claude AI, JD Edwards Orchestrator, and EnterpriseOne Classic Page.

The development team began by providing Claude with six months of batch job data in Excel. They asked the AI to recommend meaningful dashboards that could identify inefficient jobs, optimize hardware resources, and highlight execution patterns.

Through iterative prompting, the team refined the dashboard designs and established the most useful metrics. Once satisfied, they asked Claude to generate the HTML code necessary to build the dashboards and load it in a JDE-compliant format by using the UDO “Classic Page.”

The resulting architecture used the embedded Java Script connector “e1pagehelper.js” to authenticate the session without any credentials and to recall a published orchestration. Orchestrations retrieve the data from the JD Edwards database through the standard Connector/Connection components. Connections allow parametric SQL statements once expressed in Groovy language. Claude AI helped to fulfill this requirement and properly manage the information exchanged in JSON format.

Importantly, Claude was used to generate the dashboard components outside the production environment. Those components were then imported into JD Edwards, allowing the finished solution to operate internally without exposing the ERP system to the internet.

For users, the experience remained familiar. They simply selected a page in the JD Edwards homepage and accessed an interactive dashboard instead of navigating traditional batch job inquiry screens.

This separation between AI-assisted development and internal execution allowed Lucart to benefit from generative AI while maintaining control over its ERP environment.

Turning Batch Job Data into Operational Insights

The dashboards gave Lucart’s technical teams immediate visibility into several operational areas.

Batch execution efficiency: Color-coded indicators identified jobs producing zero records, exceeding expected execution times, or remaining unused for extended periods. Red highlighted jobs producing no records, yellow indicated excessive execution times, blue identified unused jobs, and green represented normal activity.

Queue utilization: Visualizations revealed uneven workloads across processing queues, making it easier to identify congestion and available capacity.

Server load balancing: Dashboards compared execution volumes across servers, helping administrators determine whether workloads needed redistribution rather than additional hardware.

Execution scheduling: Time-based views exposed periods of unusually heavy batch activity, allowing the team to identify conflicts with critical business operations.

One particularly valuable finding involved warehouse activity. Lucart discovered that excessive batch processing coincided with the early-morning period when trucks were waiting for warehouse operations to begin.

By identifying that congestion visually, the team could move appropriate jobs into less demanding time slots, potentially including evenings or weekends.

Because the dashboards refreshed quickly, administrators could adjust schedules, rerun analyses, and immediately evaluate the impact of their changes.

Measurable Improvements in the First Week

Lucart’s initial results demonstrated how much operational value was hidden within data the company already possessed.

Results reported by Lucart during the INFOCUS 2026 presentation.

The team redistributed 38% of daily batch volume to reduce interference with warehouse operations, removed more than 400 jobs that served no useful purpose, and improved load balancing after discovering that one of three servers handled approximately half the execution volume.

Bartolozzi emphasized that obsolete batch jobs frequently remain active because nobody reports when they are no longer needed. Changes to business processes, upgrades, retired customizations, and discontinued activities can leave unnecessary processing behind.

The dashboards made these previously overlooked inefficiencies visible without requiring administrators to investigate every job individually.

For Lucart, which operates JD Edwards in a private cloud using its own hardware, these improvements also supported more efficient infrastructure utilization and reduced the potential need for additional equipment.

Extending AI Beyond IT Operations

The batch monitoring project was only the beginning. Bartolozzi described plans to apply similar dashboard technology across procurement, inventory, finance, quality management, and sales.

Potential applications include monitoring slow-moving inventory, supplier performance, reconciliation volumes, quality nonconformities, and order-processing activity.

Lucart also demonstrated a different AI integration model involving manufacturing waste management.

In this application, plant controllers interact with an AI assistant through Microsoft Teams and Copilot Studio. A gateway server communicates with JD Edwards Orchestrator to retrieve production information, allowing the assistant to analyze waste levels across machines and shifts.

The system uses predefined expectations to flag suspicious results. For example, unusually low waste percentages might indicate incomplete reporting, while excessive waste could signal a production problem requiring immediate attention.

After reviewing recommendations, the plant controller can approve adjustments that are written back to JD Edwards through the gateway and Orchestrator.

Unlike the dashboard project, where AI generates components subsequently deployed inside JD Edwards, this approach allows an AI-powered interface to request ERP data and submit approved updates.

From Rule-Based Dashboards to Predictive Analytics

Lucart’s current dashboard approach primarily identifies conditions based on predefined rules. Administrators establish thresholds, and the system highlights jobs or workloads that meet those criteria.

The next objective is anomaly detection, where systems proactively alert administrators to unusual activity rather than requiring them to initiate an investigation.

Longer term, Bartolozzi envisions predictive capabilities that could evaluate proposed batch jobs before deployment, warning administrators about potential queue congestion, excessive resource consumption, or ineffective processing.

However, expanding AI adoption introduces another challenge: managing multiple agents developed internally or supplied by different vendors. Bartolozzi acknowledged that Lucart may eventually require a broader orchestration platform to coordinate these technologies without becoming locked into a restrictive ecosystem.

Practical Lessons for JD Edwards Organizations

Lucart’s experience illustrates that organizations do not necessarily need a major platform investment to begin applying AI to JD Edwards.

A practical starting point is identifying an operational problem where useful data already exists but remains difficult to interpret. Generative AI can help design dashboards and produce supporting code, while established JD Edwards capabilities handle secure data access and execution.

Equally important is governance. Bartolozzi cautioned that AI decisions carry real business consequences, making oversight a responsibility shared by IT, executive leadership, HR, and compliance teams.

The central takeaway is that AI can deliver immediate value when applied to specific operational challenges using existing ERP investments. Lucart’s approach demonstrates how organizations can move from manual analysis to actionable intelligence, improve performance, and establish a practical foundation for more advanced automation without unnecessary complexity or expense.

Want more?

Explore more content and resources to help you get the most from your JD Edwards investment:

• Visit the Quest Learn Library for blogs, how-to demos, and on-demand sessions.
• Connect with peers in one of our Quest JD Edwards Community User Groups to swap stories, ask questions, and share tips with other users facing the same challenges.
• Bookmark LearnJDE.com for updates from JD Edwards development.

Not a Quest member yet? Join today and tap into the ultimate Oracle customer network.