Part of my daily work is collating model and serial numbers information. This is to generate the description for the delivery order. Some customers want it by site, which means inventory collection. After copying the text data and pasting into the DO for a few devices, I wanted an automated way.
These devices are not yet onboarded to Solarwinds so no inventory report. Sometimes the customer may not indulge you. So I needed a program to convert the plain text output into an Excel or friendly to Excel format. After some searching, I found a program, tech2xl, on GitHub that does this.
There were a few bugs, the new interface types were causing it to crash. I poked around and got it working.
Enter stage right, ChatGPT. There are a few soft errors in the interpreter that irritated me. So I uploaded the Python program to ChatGPT. It made a few fixes and suggested some improvements. I created a project and included all of it.
Enter stage left, another up and coming network vendor. The format is different and my program can’t handle it. You can see where this is going. I can ask ChatGPT to convert plain text to Excel or I can write another program. The problem is that there is a limit on the free tier for this type of generative AI.
Enter stage right, vibe coding and Claude. Vibe coding is giving a generative AI prompts to create a program. Since I have my input files, I uploaded one and told it to generate the Python code that converts them to an Excel-friendly format. It works! I did a second round to fine-tune and handle no-input errors. I ran the code through PyCharm and compiled it as a standalone program. All this on a Sunday afternoon. If I had done this from scratch, it would have taken me months, as I am not a Python programmer.
Anything that needs to be done twice, script it.