.png)
Building a report in Oracle BI Publisher has traditionally required a lot of manual work. Developers need to identify the right Oracle tables and views, determine the joins, write and troubleshoot SQL, build the data model, and then create the RTF layout.
AI can change that process, but it does not eliminate the need for Oracle expertise. Instead, the 80/20 method uses AI to create much of the initial Oracle BI Publisher report while leaving validation, testing, and refinement in the hands of the person who understands the business requirements and Oracle environment.
Want to see the 80/20 method in action? Watch the full video walkthrough to see the process from the initial prompt through the finished Oracle BI Publisher report.
What Is the 80/20 Method for Oracle BI Publisher?
The idea is simple: AI drafts the 80%, and you own the 20%.
Rather than manually writing everything from the beginning, users can describe the Oracle BI Publisher report they need using business context and the appropriate technical details. AI can draft the SQL and, later, help build the RTF layout. The user is still responsible for validating the output and correcting it for their specific Oracle Fusion environment.
That validation matters. AI does not automatically know your Oracle release, configurations, or data. It may also make incorrect assumptions, including using Oracle E-Business Suite tables when the report is intended for Oracle Fusion.
The quality of that initial 80% depends heavily on the information AI receives, which makes building the right prompt an important part of the process.
Prefer to follow along in video format? Watch the full walkthrough of the 80/20 method, covering the same process from building the initial prompt to testing and refining the final Oracle BI Publisher report.
Demo: Building Oracle BI Publisher Reports with ChatGPT
WATCH NOW9 Elements of a Strong Oracle BI Publisher Prompt
A strong Oracle BI Publisher prompt gives AI both the business context and technical requirements it needs to build a useful starting point. The 80/20 method includes nine key elements:
- Role and context: Tell AI the role it should take, such as a Fusion BI Publisher developer, and clearly specify whether you are working in Oracle Fusion or EBS.
- Business question: Explain the decision the report needs to support so AI understands what matters.
- Tables: Provide the tables you already know. If you do not know where a field comes from, describe what you need and ask AI to identify the table.
- Output columns: Clearly define the columns the report should return and the order they should appear.
- Parameters: Specify the parameters using the syntax BI Publisher expects.
- Business rules: Define requirements such as which statuses to include, which lines to exclude, and how currency should be handled.
- Grouping and sorting: Explain how the results should be grouped, totaled, and ordered.
- Technical constraints: Include requirements such as using Oracle SQL and avoiding a trailing semicolon that would cause the report to fail.
- Assumptions: Ask AI to list every assumption it made so you know exactly what needs to be validated.
Only some of these elements are highly technical. Much of a strong prompt comes down to clearly communicating the business context, requirements, and expected result.
The final element is especially useful. Asking AI to list its assumptions creates a checklist that can be used when testing the report. Instead of guessing what AI assumed, you can work through those assumptions and validate them against your actual Oracle Fusion environment.
Weak Prompt vs. Strong Prompt
The impact of that additional context becomes clear when comparing two prompts for the same Oracle BI Publisher report.
In the example, the goal was to create a purchase order spend report showing PO spend by supplier, broken out by business unit, category, and buyer for a selected date range.
A simple prompt asking AI to “write me a SQL query for PO spend by supplier in Oracle” left too much open to interpretation. AI assumed Oracle E-Business Suite, used an EBS supplier table, omitted requested groupings, and included a trailing semicolon that Oracle BI Publisher would reject.
The stronger prompt applied the nine elements above. It specified Oracle Fusion, the business goal, known tables, required fields, output columns, parameters, business rules, grouping, technical constraints, and a request for assumptions.
The resulting SQL was much closer to the desired report, although testing still uncovered an issue with the date parameter that needed to be corrected for the actual environment. After that adjustment, the report returned data grouped by business unit, supplier, category, and buyer, along with PO count and total PO spend.
That is the 80/20 method in practice: AI can accelerate the initial build, but the user still owns testing and validation.
Building an Oracle BI Publisher RTF Layout with AI
The same approach can extend beyond SQL to the Oracle BI Publisher RTF layout.
In the example, sample XML data was first masked so that the values were replaced while the XML structure remained intact. AI was then given the masked XML along with information about the data types, desired layout, and report requirements.
The first RTF looked complete but contained problems. Only one supplier row displayed even though the subtotal contained the full amount, the grand total showed zero, and the template included a note that had not been requested.
Instead of simply telling AI that the report was wrong, the next prompt explained each issue and what the correct result should look like. After those corrections, the report displayed each supplier along with the appropriate business unit subtotal and grand total.
The lesson was the same as with the SQL: be specific about what you need, test what AI produces, and clearly explain what needs to change.
What’s Next?
The 80/20 method shows how AI can support Oracle BI Publisher development without removing the expertise and judgment required to build an accurate report. Give AI context rather than just a request, be specific about what you know, review its assumptions, test the output, and own the final 20%.
As AI becomes a bigger part of how organizations work within Oracle Fusion Cloud, knowing where and how to apply it will be increasingly important. Elire can help your organization explore practical AI opportunities across Oracle Fusion Cloud while keeping your business requirements and processes at the center. Connect with Elire to start the conversation.
Want to stay current on Oracle AI and Fusion Cloud? Subscribe to Elire’s Cloud Newsletter for the latest insights, updates, and practical guidance.
You can also follow Elire on LinkedIn and subscribe to our YouTube channel for Oracle insights, product updates, demonstrations, and more.
