With the goal of delving as deeply as possible and uncovering the most advanced potential and features of ChatGPT, I embarked on my own creative ChatGPT research. Utilizing a design thinking approach and armed with my creative skills and full of questions, I dove into this process of a lot of testing and interesting experimentation. I had the opportunity to discover new concepts, interesting insights, and even developed a new promising prompt engineering technique with great potential for optimizing the development process and reducing development costs. Below, I will share about these new concepts and techniques.

It is a new and promising prompt engineering technique with the groundbreaking capability to enable ChatGPT to handle and process multiple prompts with different instructions simultaneously, in a single prompt input. The Concatenated Prompt Technique (CTP) streamlines interaction with AI by merging complex commands, thus enhancing efficiency, reducing API calls, and cutting down on development costs. This approach also facilitates deeper understanding by the AI, allowing developers to provide much deeper and detailed contextual information to the model. It also permits detailed and sequential control over the model behavior, enabling the model to handle much more complex and intricate interactions resembling app or website functionality (Prompt App).

A “Full Prompt” is a simple yet important concept that describes the optimal state of a prompt, where the prompt is complete and filled with all the information needed by the AI to address the request precisely by also providing detailed output format instructions as well. Understanding this concept means understanding that if you are not obtaining the necessary quality of response from ChatGPT, this can be an indicator that your prompt has not yet reached its “full prompt” state.

It’s a very interesting concept that aims to leverage the use of ChatGPT from a simple and limited text environment to a more rich and immersive experience. By utilizing advanced markdown with precise output instructions, it allows creative professionals to craft more refined chat experiences, going beyond simple text interactions.

A “promptapp” is a unique concept referring to a new class of applications created solely through prompts designed to run within the ChatGPT environment, treating ChatGPT as both the platform and the interpreter for these apps. This approach transforms the initial prompt into the “code” of the application, leveraging the capabilities of language models to process and execute complex interactions. Essentially, it’s a way to develop applications that operate entirely within the confines of a chat interface, without needing traditional programming languages for user-level interactions.

My research continues, and there are already some new developments in the pipeline. Some of the research projects listed below are in the early stages, while others have already been completed, and I will publish them here soon. So, at this point, I’ll just mention a few more interesting things to come

This is actually a prompt with Bulk Image Generation capability. It is totally free because it runs with ChatGPT 3.5 and Pollinations AI. Yet, it is a robust and powerful bulk image generator with a unique feature that allows users to generate multiple images simultaneously, while also allowing users to adjust parameters of each image differently within the same generation request. For example, you can generate an image of character X, performing action X in scenario X, and ask the model to ensure that each image of the generation will have everything the same and will change only the action on each image, allowing the user to generate the same image while varying only the action performed by the character in each image. A super simple and easy way to perform various tests in image generation by AI. Comming soon…

It is a very promising technique that uses content from mixed sources to allow the creation of applications that offer users a high degree of customization in generating prompts for various purposes, such as image prompts, text-to-video prompts, text-to-audio prompts, and so on. At the same time, it gives developers total and granular control over all other parameters of creation, ensuring that the result generated by the model is always in line with the project scope and also maintains all the necessary quality parameters to generate results that meet user expectations. All of this in a single input prompt. Comming soon…

In this world of prompt engineering in ChatGPT, placeholders play a fundamental role in the model’s ability to understand our instructions in a more logical and structured way. Through placeholders, we can create clusters of logical arguments. So, once these clusters are connected, it’s possible to create a logic chain in the model’s memory, a set of logical instructions that complement each other and form a kind of logical puzzle. I believe that expanding the use of this approach, prompt designers will achieve a higher level of control over ChatGPT’s behavior and answers, thus allowing for a much more efficient use of the models and providing professionals in the field with more resources to instruct their models with greater precision.
Work in progress…

I’m now looking for an interesting AI project to put these resources into practice, or any company or institution wanting to support and invest in development and research with ChatGPT. If you want to elevate the use of ChatGPT and put your project at the forefront of ChatGPT usage, then let’s join forces and move forward. Please, get in touch…