Over the weekend I made my first GPT.

It’s an IBP GPT, maybe the world’s first. Although you can make GPT’s more sophisticated, you can start rather simple. That’s what I did.

I took an open AI membership, created a GPT, wrote a prompt, wrote four prompt indicators. Told it to be friendly and gave it some other guidance. My aim was to help an IBP user brainstorm some solutions to a typical problem during an IBP cycle.

The IBP GPT will ask for some context of a problem and give solution directions. After the user chooses a solution direction the GPT will dig a little bit deeper to provide ideas on how to go about solving it. The GPT will provide an example of the chosen solution direction. The GPT will end by providing some tips to improve decision quality and say “I hope this helps you improve your IBP recommendation”

My first impression after using it about 10 times, including to “close a $200,000 NSV gap in my demand plan versus budget”:

• It has useful ideas, although they’re generic
• Most ideas I knew about, but some I hadn’t thought about
• It’s a bit elaborate, maybe too many ideas, not all relevant
• How do I really implement this stuff? I might have to ask that as well!
• The decision quality tips are poor (that requires better prompting in the GPT)

It is certainly not perfect. It seems best for brainstorming ideas. I can further tune the prompt or tell it to exclude certain advice that is out of scope in a common IBP cycle. I might not want advice to change my technology platform to solve a problem in my monthly IBP cycle. I would love you to test it and provide some feedback so I can maybe improve the IBP GPT.

Here is the link: https://chatgpt.com/g/g-ZNps8xRTP-ibp-recommendation-guide

OpenAI had to train it’s GPT. It’s relatively old data, not real time. But simply imagine a GPT like this on top of your latest plans and financials, all your historical company data and some external market indicators relevant to make monthly IBP decisions.

You’ll understand that we’ve entered a new era