ChatGPT & Me Episode 2: A shared understanding…sort of
Documenting workflows can be tedious. If ChatGPT can become a flowchart SME — producing good workflow diagrams from input such as photos of sticky notes or whiteboards — it has the potential for helping a team or a department reach its workflow clarity goals quickly.
In my previous blog (ChatGPT & Me Episode 1: Training the Trainer), I found all sorts of ways to communicate requests to ChatGPT and not obtain the results I expected.
For Episode 2, I started with the question: Are you able to generate a sample flowchart diagram using the ANSI/ISO Standard Shapes for flowcharting?
The results are shown below.
Training AI to be an effective workflow assistant is as much about identifying the parameters of its capabilities as it is learning to write effective prompts. This training episode taught me that ChatGPT can create workflow diagram code, which can be transferred to Python for generation of a diagram.
I started with the question: Are you able to generate a sample flowchart diagram using the ANSI/ISO Standard Shapes for flowcharting?
ChatGPT’s first response was to provide a Graphviz script, instead of the diagram image I expected.
As we conversed, I learned ChatGPT wasn’t able to provide a diagram, but it could attempt to create working links to PDFs and .png images (all of which failed to open).
ChatGPT had mentioned Python in its previous response.
Out of curiosity, I selected the Python GPT from ChatGPT’s GPT library. Python — I just learned — is a programming language.
I asked it to convert the code to a flowchart.
After I pasted the Graphviz code into the Python chat box, Python created the flowchart diagram from the code.
Operational excellence begins with clarity on workflows and procedures.
This blog was co-written by a human, ChatGPT, and Python GPT and edited by a human!
© 2025 Lori K. Barbeau