3.0 Bottom Line Up Front/TL;DR

Develop and share advanced AI user and integrator skills.
Demonstrate that ChatGPT, through its API, can be used to develop a custom AI application using custom instructions and taxonomy.
Demonstrate that an AI application can be integrated with a Content Management System to form AI as a Service (AIaaS).
Push the envelope on using generative AI for research, development, marketing development, and training.
Identify a common "concept" on approaching AI to maximize production and safeguard property.

AI as a Service - an application that is built upon a generative AI engine, with custom instructions and taxonomy with a web interface
Turbo vibe coding - using two different Gen AI services for hard problems for approach diversity, to catch obscure errors, and to achieve a solution. An application of turbo coding from communications theory.
Rabbit holes - when Gen AI takes you down a development path, only to find at the end that it is non-viable or non-optimal
Donuting - when Gen AI takes a circular path, trying multiple steps again and again, convinced that they will eventually work

Created an AI Service with a branded, webhosted AI application and sandboxed user access
Developed new concepts in managing AI, such as AI as a consultant
Embodied an application in a portable taxonomy which can be used with other models and agents

ChatGPT and Claude enabled me to finish this in record time. In two months, I accomplished what would have taken me over a year to complete otherwise. In addition to increased productivity, ChatGPT helped me with marketing, strategy, brand research, etc.
Treating ChatGPT and Claude as consultants allowed me to manage their strengths and weaknesses. I assumed that they were holding back information and, to some degree, could become a leak of competitive information.
ChatGPT and Claude have software development limitations which can be tied to the type of task. Both are really good at mainstream programming languages such as Python and PHP. When the coding task is state of the art, obscure, or lightly used, both LLM's are limited in ability and success. Examples include Android development in Kotlin and Jetpack Compose toolkit, template development for Joomla.
Human software developers still have a place. Possible 20-25% of this project was developing the web style that you see now. It was painful. For the final debugging, neither ChatGPT nor Claude could find the error. Ultimately, I fixed the problem.
A key skill for junior software developers will be managing AI agents and software development. Continuing with the consultant analogy, junior developers will manage the consultants.

[This article's text was generated without AI. Graphics were AI generated.]