AI Experiment 1
The Ultimate Product Owner Guide
Is it possible to build an Ultimate Product Owner Guide by using OpenAI’s GPT Builder, and pass the Product Owner Open?
A couple of weeks ago, we were trying out the new GPT Builder of OpenAI. After being on the waitlist for what felt like months, we could finally try it out, and it was mostly a great experience. In this first experiment, we’ll tell about one of the first GPT’s we built and tried. More experiments with more GPT’s are likely to come!
First of all: The simplicity of the interface, in combination with how well the actions responded to our input was amazing. The flow that OpenAI follows on helping a user to build their own GPT, with a simple, open question like: '“What would you like to make?” and further “How does this one look to you?”.
As you can see in the images, we wanted to build a GPT that would help Product Owners in their day-to-day job. In a nutshell, for whoever doesn’t know this trendy job title of Product Owner: A product owner collaborates with stakeholders and the development team to define and prioritize features, manage the product backlog, and align the product vision with business goals. This role requires business, tech, communication, and leadership skills.
Why make a GPT for Product Owners you might ask? We’re both Product Owners and we could use a little help now and then!
Now we won’t bore you with the complete step-by-step approach and all the different prompts we’ve tried. However, it's worth noting that there is an endless array of try-outs available for you to further instruct your GPT. These trials will allow you to refine its responses to align precisely with your preferences, making the process more customizable and tailored to your specific needs.
The proof of the pudding is in the eating, so we needed to try and see if this Product Ally is giving sound advice. And what better way than to let it do an actual exam? We’ve tried with multiple prompts, in different ranges of success. Our first attempt the Ultimate Product Owner Guide, as it was called by then, was already a promising 65%.
With more tweaking, adding different sources, and so on, we had 12/15 correct. Further fine-tuning made even the last, previously incorrect answers correct. Now our GPT has a 14/15 score in the demo exams!
Should we do a real exam to find out? Let us know in the comments.
Want to try the Ultimate Product Owner Guide for yourself? Click here.
TL;DR: We, as Product Owners, built a GPT that outscores us on demo exams… ✅
