So in this scenario I decided to use one of the pre-existing data sets I had to see what Chat GPT would tell me . I am probably going to be phasing out ChatGPT over the next while but will keep on using Co-Pilot. I decided to give AI the following prompt.
I was experimenting with this prompt framework during the week and found it quite helpful and insightful – the BRIDGE framework. You can read more about it down below
I have written a prompt and I have a fake dataset which I am going to load into ChatGPT to see what it gives me.
This was the final output – which is actually pretty good. In the instructions, it did specify using pivot tables but that’s not what is in the Excel files.
This is a structured, six-part methodology for crafting effective, high-quality prompts for large language models (LLMs), often used in business contexts to ensure accuracy and relevance.
- Background: Provide context for the request.
- Request: Clearly define what you need the AI to do.
- Inputs: Provide the necessary data or constraints.
- Deliverables: Specify the format of the output (e.g., report, code, table).
- Guardrails: Define limitations (e.g., “do not mention X”).
- Evaluation: Set criteria for success.