By now, you’ve likely experimented a bit with ChatGPT. I’ve extensively discussed it in my Newsletter and shared in-depth insights on my LinkedIn feed and blog. If you haven’t done so yet, you can start by reading the ‘ChatGPT for Oil & Gas’ series, with part 1 available here.
Today, we’re going to explore CustomGPTs, which are essentially the same as regular ChatGPT but tailored to a specific subject of your choosing.
And the best part? You don’t need any coding skills. Within just a few minutes, the ChatGPT interface allows you to create a bot designed to engage in conversations on the particular topic you’re interested in, making every interaction more relevant, insightful, and engaging for its users.
Whether for educational purposes, customer support, or niche hobbies, CustomGPTs open up a whole new world of possibilities.
To demonstrate how it works, I have created a ‘PRMS Advisor’.
As you may know, the PRMS Guidelines are outlined and consolidated in a few PDF documents, the key ones being the 2007 PRMS foundational PDF, the 2011 PDF with extended guidelines and examples, and the 2018 Update. They are all publicly available for download from the SPE website at https://www.spe.org/en/industry/reserves/.
If you have ever been involved in a Resource and Reserves classification process, you most likely will need to consult these guidelines, and often, tracking down those elusive PDFs can be a hassle.
So, I thought this would be a good use case to have this advisor readily available in your pocket, where one can ask specific questions and have the bot scan through the content of the guidelines and return the most relevant answers.
Depending on public response, I may create a video tutorial to demonstrate the step-by-step process of creating the PRMS Advisor. Stay tuned, but for now, let me outline the key steps involved in creating a CustomGPT:
- Identify Your Topic: Decide on the specific subject or area you want your Custom GPT to focus on. This could be anything from cooking recipes to financial advice. In my case, the topic is clear: the PRMS Guidelines for the Oil & Gas Resources and Reserves classification process.
- Gather Your Data: Collect relevant texts, articles, or any other material that covers your chosen topic comprehensively. This data will serve as the foundation for training your Custom GPT. For training our PRMS Advisor bot, the key documents to use are the PDFs from 2007, 2011, and 2018.
- Prepare Your Data: Organize and clean your collected data to ensure it’s ready for processing. This step might involve removing irrelevant information, correcting errors, and formatting the data consistently. There is nothing to do here, as we will rely on the PRMS PDFs as they are.
- Train, Test and Refine: Test your Custom GPT with questions or prompts related to your topic to see how well it performs. Based on the responses, you might need to refine your model by further training it with more data or adjusting its parameters. This step is where your expertise comes into play. Pose the challenging questions and refine your instructions iteratively until the answers are satisfactory. This is also the stage where you provide guidance on how to respond to off-topic questions, among others.
- Deploy: Once satisfied with your Custom GPT’s performance, publish it publicly so your users can interact with it.
- Monitor and Update: Keep an eye on how your Custom GPT is performing with real users and be prepared to make updates. You might need to retrain your model with new data or tweak it to maintain its accuracy and relevance over time.
Here’s how it appears: you’ll see a small logo or image, a title and description, along with a set of pre-made questions that serve as examples of the types of queries users might wish to ask. From there, users simply start conversing with the bot in plain English.