I purposely avoid getting too far into the technical details of Information Technology on this blog, but this brief discussion is both important and worth your time. The significance of Artificial Intelligence (AI) in ERP systems, particularly within the context of People, Ideas & Objects Preliminary Specification, is profound. Larry Ellison’s Oracle AI World Keynote address contained subtle points that, I believe, require an understanding of his perspective to fully grasp. When it comes to databases, Larry Ellison stands apart, much like Steve Jobs or Elon Musk in their respective fields. 
His presentation began by addressing AI training and reasoning, specifically highlighting the role of private, structured data stored in Oracle’s global databases. This structured data adheres to specific principles, which are the focus of this post.
Ellison then asserted that “AI can Perceive, Understand & Reason across all types of data.” This capability is mirrored by Palantir’s software, which excels in handling private unstructured data. The key distinction between structured and unstructured private data lies in the normalization process and the relational nature of the data stored in Oracle AI Database 26ai. Commingling structured and unstructured data, as Palantir does, raises concerns about consistency and reliability. Database management systems are typically evaluated based on their performance in implementing the “relational model.” Oracle has historically been a leader in this area, and we will observe the impact of these new developments.
Interestingly, upon reading the relational model, Ellison was inspired to create the first relational database, leading him to co-found Oracle in 1977. According to ChatGPT, Larry Ellison, along with Bob Miner and Ed Oates, established Software Development Laboratories (SDL) in 1977, which later became Oracle Corporation. In 1979, they launched Oracle version 2, the pioneering commercially available relational database management system (RDBMS) that used Structured Query Language (SQL). This innovation was influenced by Edgar F. Codd’s seminal 1970 paper on the relational model of data.
An annotated version of Dr. Codd’s paper, relevant to this blog post, is available here. I strongly recommend a thorough review and understanding of this paper. After 55 years, its foundational principles for data management remain highly relevant. Today, relational databases appear to be the most valuable assets for a firm. C. J. Date’s textbook, Database Design and Relational Theory, also offers excellent guidance here. For those looking to effectively leverage AI and IT, proficiency in relational databases will provide a significant competitive advantage.
In the abstract of Dr. Codd’s paper, a crucial word stands out: “inference.” A properly normalized relational database allows users to “infer” information from its data. This concept is elaborated in section 2, “Redundancy and Consistency,” specifically subsection 2.1, “Operations on Relations.”
We will now explore the implications of these developments for the oil and gas industry, building on the Preliminary Specification. A clarification is necessary before proceeding: we are cautious about exposing data in ways that compromise security, accuracy, or reliability. Our user community and service providers—the accounting and systems providers for producers—have been empowered to leverage the numerous benefits offered by the Preliminary Specification. Producers will lack the in-house accounting knowledge or resources due to the reorganization from the Preliminary Specification.  However, service providers will possess the necessary background, system understanding and authority to furnish producers with the accurate, factual data they need. AI generating “garbage out” is not in the best interest of our organizations, the industry, or the producers.
People, Ideas & Objects must finalize our data model in light of these significant advancements. AI has already assisted us in developing it, and we anticipate continued and expanded use of this technology. While I am hesitant to fully embrace Oracle’s enthusiastic promotion of APEX, ChatGPT indicates its use of programming languages as follows:
Oracle APEX application code generators primarily use PL/SQL for server-side logic and JavaScript for client-side functionality. APEX itself is built on the Oracle Database, so the generated processes, triggers, and dynamic actions are mostly executed using PL/SQL, with HTML, CSS, and JavaScript for the user interface.
Oracle’s choice to “primarily rely on PL/SQL for server side” is brilliant. It is their proprietary, procedural language that allows for conditions and loops to be embedded close to the database. Having used it in our developments in the early 1990s, I find this choice entirely appropriate. The “primary” qualification suggests that Java can be injected for handling database triggers and, I presume, stored procedures, or if not Java, then PL/SQL. The use of client-side technologies is less critical to this discussion. Other relevant considerations include the ongoing optimization of Java for AI, and the fact that SQL and PL/SQL are currently running on NVIDIA GPUs. (Oracle proof of concept.)
I largely view APEX as a client-side access tool. For individuals well-versed in the requirements of Oracle AI Database 26ai, with authorized access to People, Ideas & Objects, our data model, and the Preliminary Specification, this would be immensely beneficial to oil and gas producers. It would support both daily ERP operations—identifying profitable areas and opportunities for growth—and AI-driven inferences on the database.
We do not need to delve into vectoring, its storage, or use at this stage; these are features we will implement later. For now, our transactional needs necessitate the structural integrity of the Oracle relational database.
Regardless of how this technology is marketed by any vendor, the core point remains: Oracle is very close to fully meeting the relational model requirements defined by Dr. Codd in 1970. Our diligence in developing and implementing AI over the next decade will be a top priority. With Oracle, we have access to research, development, unparalleled tools, products, and support. We must also balance these technological advantages with our objective of providing producers with the most profitable means of oil and gas production.
Implementation
Oracle APEX and its Agentic AI have a significant role to play in the Preliminary Specification. Our user community and service providers would find immense value in its application to support profitable North American oil and gas producers. With exclusive licensed support from People, Ideas & Objects developers, they would be able to facilitate the dynamism and innovation that producers are engaged in, through detailed, accurate ad-hoc reporting and other essential functions.
Oracle APEX operates with two SQL database sublanguages: Data Manipulation Language (DML) and Data Definition Language (DDL). Responsible database developers will never grant general access to DDL, leaving our user community and service providers with DML access only to the fields within the tables they are authorized to view.
Oracle APEX is a powerful tool, and Agentic AI currently benefits from a robust development process where Oracle verifies the safety and validity of each Agent before release. This makes it an incredibly potent tool for industry professionals who utilize the Preliminary Specification, our user community, and service providers. We do not anticipate extensive hands-on development use by producers themselves, as this involves accounting information that requires appropriate preparation for presentation.
However, as an ERP system, we are involved in conducting mission-critical transactions. APEX cannot be involved in the development, creation, or management of these transactions. Processes such as the Material Balance Report have persistent, asynchronous, and stateful issues. APEX is stateless and, therefore, currently incapable of handling such demands. In terms of transaction processing and process management, People, Ideas & Objects’ software development requirements exceed APEX’s current capabilities. Java, SQL, and PL/SQL will be essential in these areas. 
Until proven otherwise, we must take all necessary measures to ensure the security, reliability, and integrity of producer data is captured and reported accurately. We look forward to fulfilling our objective of ensuring a renewed culture of reserves preservation, performance and profitability is established. With Oracle we know we can. Supporting producers, and the service industry as they undertake the next 25 years in oil & gas which have to be its most challenging ever.