It’s rewarding to set a technical goal and build a solution that works. This project started as a manual task and evolved into an automated workflow that saves time and improves consistency.

The Challenge: Converting Links for Our LCMS

The task was to convert a list of topic-related links into structured, clickable XML blocks that our Learning Content Management System (LCMS) could import. The goal was to generate consistent, import-ready hyperlinks embedded in properly structured XML, without manually formatting each one.

Starting with Python

Although I’m not a developer, ChatGPT helped guide me through the process.

I started with a Python script to process Word documents and spreadsheets. The goal was to extract link titles and URLs, then format them into XML.

I experimented with a few techniques:

• Parsing Word paragraph runs
• Extracting hyperlinks using .rels files

However, Word’s internal structure was too inconsistent. I switched to a more reliable source: a spreadsheet with two columns, one for link titles and one for URLs.

Parsing the spreadsheet row by row simplified the process. With Python, I could generate clean XML output consistently.

Finding the Right XML Format for our LCMS

Our LCMS requires a specific hyperlink structure. Early attempts using tags such as <a> or <Link> didn’t work. After some trial and error, I identified the correct format.

With the correct structure in place, the Python script could generate well-formed XML blocks with the required elements and attributes.

Automating the Workflow with a Custom GPT

Although the Python script worked, I wanted a scalable, no-code solution that my team could use.

Using the ChatGPT builder, I created a custom GPT that:

• Accepts a spreadsheet upload
• Generates unique GUIDs
• Formats each row into the correct structure
• Outputs a ready-to-import XML file

Now, instead of running a script or writing code, I upload the spreadsheet. The GPT returns the XML in seconds. It’s fast, reliable, and repeatable.

Results

By moving from a manual, script-based approach to an automated workflow powered by a custom GPT, I:

• Saved hours of repetitive formatting work
• Ensured consistency across XML imports
• Enabled team members to generate XML without writing code

This solution demonstrates how combining Python, XML, and tools like GPTs can streamline even highly specific technical workflows. For teams working with structured content systems like our LCMS, automation can greatly improve efficiency.

A Few Caveats

Of course, there’s more work ahead. Some challenges remain:

• All links across all topics must be reviewed and updated.
• A future transition in our delivery mechanism might require more changes.

These are open questions, and the long-term solution might look different. But this approach already provides something valuable. It offers a practical solution and new insight into how we can scale and improve our documentation workflows.