What happens when experienced business analysts leave an organization, requirements become increasingly ambiguous, and developers spend more time investigating problems than writing code? For Dealer Tire, the answer was to rethink its JD Edwards development process using artificial intelligence.

During his INFOCUS 2026 session, From Ambiguous to Actionable: How Dealer Tire Uses Claude AI to Transform JDE Development, Joe Davis shared how Dealer Tire integrated Anthropic’s Claude AI into its JD Edwards software development lifecycle. The result is an evolving solution that improves requirements gathering, accelerates testing, simplifies technical analysis, and helps preserve institutional knowledge.

Rather than positioning AI as a replacement for developers or business analysts, Davis demonstrated how it can serve as a practical accelerator, allowing technical teams to focus on higher-value work while reducing repetitive tasks and unnecessary development cycles.

The Challenge: Ambiguous Requirements and Development Rework

Dealer Tire’s transformation began with a familiar problem: inconsistent requirements and gaps in business knowledge were creating unnecessary work for developers.

When experienced business analysts departed, their replacements often lacked the JD Edwards expertise necessary to answer technical questions or fully investigate incoming requests. Developers consequently became responsible for interpreting requirements, researching business processes, and determining what users actually needed.

This created a cycle of development, testing, clarification, and rework. Davis explained that some tickets experienced as many as a dozen rounds of revisions before reaching an acceptable solution.

Dealer Tire established a straightforward objective: provide consistent inputs so developers can concentrate on development.

When company leadership introduced an AI-first engineering initiative in March 2026, the team saw an opportunity to address these challenges. Despite receiving broader access to Claude licenses only in mid-April, Dealer Tire developed its initial proof of concept in approximately 45–60 days, including time spent addressing security and compliance requirements.

Connecting Claude AI to JD Edwards Through MCP

Dealer Tire initially experimented with a manual approach. Developers exported event rules from the JD Edwards fat client and supplied them to Claude alongside Jira tickets and acceptance criteria. Although useful, the process required substantial manual effort to provide the necessary technical context.

The team subsequently developed a local Python-based Model Context Protocol (MCP) server connecting Claude Desktop to JD Edwards through existing Oracle REST APIs.

The integration provides 14 tools supporting activities such as authentication, retrieving table definitions, examining event rules, identifying orchestrations, and querying object specifications. Through natural-language prompts, developers can investigate JD Edwards artifacts without manually navigating numerous fat-client screens.

The solution brings together information from several sources, including UBE event rules, business views, table specifications, UDOs, orchestrations, Jira work items, and internal documentation.

Importantly, the implementation incorporates strict security controls. Access is read-only and restricted to nonproduction environments for JD Edwards queries. Claude uses each employee’s existing credentials and security permissions rather than a separate privileged account. Credentials are stored through a credential manager, and the team worked with information security and compliance stakeholders to review data handling and contractual protections.

Although leadership also wanted JD Edwards source code available in GitLab for broader visibility into changes, Davis acknowledged that this integration remained an unresolved objective.

A Real-World Example: Finding the Root Cause of an Ordering Problem

One of the presentation’s most instructive examples involved an ordering issue affecting Dealer Tire’s maintenance products.

A ServiceNow ticket reported that certain SKUs should be ordered in minimum quantities of 100 and multiples of 100. Instead, the ordering interface displayed increments of 99, forcing customers to order 198 units to satisfy the minimum requirement.

Initially, the request appeared to describe an application defect. However, using Claude to refine the requirements and investigate the underlying configuration revealed a more complicated issue.

A JD Edwards user-defined code contained the numeric value 99 but had a description specifying orders in lots of 100. The web application interpreted the code itself as the ordering quantity rather than using the intended business meaning.

Further investigation revealed that an employee had been unfamiliar with Dealer Tire’s separate minimum order requirements table. Ultimately, the underlying issue involved business data setup, with the web application’s interpretation also contributing to the confusion.

This distinction mattered because the original ticket did not adequately identify the source of the problem or the appropriate team to resolve it.

Davis demonstrated how Claude helped organize the investigation and clarify the requirements before additional development work began.

His preferred ticket structure starts with a problem statement of three sentences or fewer, followed by acceptance criteria and supporting context. The objective is to explain what the business needs and why, rather than prematurely prescribing a technical solution.

Generating Event Rules, Technical Analysis, and Test Plans

Dealer Tire’s MCP integration also enables developers to investigate JD Edwards applications using straightforward conversational prompts.

During the demonstration, Davis queried the company’s custom minimum order requirements application, P55MINOR, and asked Claude to retrieve its event rules. Claude identified the appropriate tools, returned the technical information, and produced a text representation resembling the event rules developers ordinarily review through the fat client.

The same approach supports cross-reference analysis, helping developers identify objects that interact with a particular application. Developers can also request pseudocode for proposed modifications or ask Claude to evaluate multiple solutions, including their advantages, disadvantages, and trade-offs.

One of the most valuable applications is automated test plan generation.

Davis demonstrated how a simple request could produce a formatted Excel workbook containing test cases and instructions for business analysts. Tasks that previously required days, or sometimes weeks, could be reduced to minutes.

This is especially useful when business analysts lack extensive familiarity with a particular JD Edwards application. Instead of developing testing scenarios entirely from scratch, they receive a structured starting point that developers and analysts can validate and refine.

However, Davis emphasized that AI-generated outputs must never be accepted without review. Even convincing technical documentation can contain errors or unsupported assumptions. Human verification remains essential.

Beyond Development: Measurable CNC Productivity Gains

The benefits extend beyond application development into CNC administration and troubleshooting.

Davis reported that Dealer Tire’s CNC team had saved approximately 144 working hours through its AI-assisted workflows. One particularly compelling example involved preparation for JD Edwards certification ahead of future Tools and Applications upgrades. A process that ordinarily required approximately 40 hours was completed in 12.

Another example involved a production print server outage that prevented distribution centers from printing packing slips and shipping tires.

The incident brought together approximately 20 employees across infrastructure, networking, development, and leadership during a troubleshooting call lasting three hours.

After reviewing various system logs without identifying the cause, the team supplied the information to Claude. The analysis helped identify a corrupted PDF repeatedly being submitted through the print queue. Once the team stopped the queue and removed the corrupted file, printing resumed.

Although the MCP integration itself remained restricted to nonproduction JD Edwards access, this separate use of Claude for log analysis demonstrated its broader troubleshooting potential.

Building a Sustainable AI-Enabled Development Process

Dealer Tire’s experience also highlighted the organizational work necessary to make AI adoption successful.

Davis stressed that executive support is critical when introducing standardized development practices. Without leadership reinforcing expectations, engineering teams can struggle to establish consistent requirements, documentation, and intake procedures.

He also encouraged organizations to start small rather than waiting for a complete integration.

A team can begin by exporting JD Edwards event rules, pairing them with an existing development ticket, and asking an approved AI tool to analyze requirements or generate test cases. Results can then be evaluated and improved through successive prompts.

Over time, organizations can expand access to internal documentation and technical artifacts, making AI-assisted analysis more useful.

Dealer Tire’s longer-term vision includes transforming its locally hosted integration into a Claude connector available to other authorized JD Edwards users. This could extend technical analysis and self-service capabilities beyond the engineering team, reducing dependence on developers for routine questions.

The company is also evaluating whether continued internal development or purchasing an established solution offers the better long-term investment. Both approaches involve costs, whether measured in licensing expenses, engineering time, or competing priorities.

The Future of AI in JD Edwards Development

Dealer Tire’s experience demonstrates that AI’s greatest immediate value may not come from automatically writing code. Instead, it comes from improving the quality of information developers receive, making complex systems easier to investigate, and reducing the time spent on repetitive analysis and documentation.

Davis also advocated configuring AI to challenge assumptions, explain its reasoning, and provide supporting evidence rather than simply agreeing with users. These practices reinforce the importance of treating AI as an analytical partner whose recommendations still require professional judgment.

For JD Edwards organizations facing knowledge gaps, inconsistent requirements, and growing development backlogs, Dealer Tire offers a practical example of how to begin.

The underlying lesson is that successful AI adoption requires more than technology. It depends on clear requirements, appropriate security controls, reliable context, leadership support, and knowledgeable people who can evaluate the results.

By combining those elements, Dealer Tire is turning ambiguous requests into actionable development work while helping its teams accomplish more without reducing headcount.

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.