"It saves us a lot of time. It's crazy," Stijn notes. "Everyone will save time. So it does impact on everyone in the Metaal union, which is a good thing."
The challenge
In Belgium, like many countries, union delegates have the right to take union leave for training, education, or executive committee meetings. Processing these requests using a time-consuming manual process at ABVV-Metaal. Stijn Merckx, of ABVV-Metaal explains: “We used fill out the employer, the name, the dates, and then I had to sign it by hand and then we scanned it and then we sent it by letter. So it was a really old and complex thing”.
For each delegate attending an event, staff had to:
Fill out a paper document with employer details, delegate name, and dates
Sign the document by hand
Scan the signed document
Send it by email to the employer
With many delegates across the region, this repetitive administrative task was took up a lot of staff time – time that could have been spent on more valuable union work supporting members.
The union needed a way to modernise this process while meeting the formal requirements of the leave application system.
The solution
After attending an AI training session, Stijn decided to tackle the problem on a Saturday afternoon. "I asked AI, maybe can I do something like that? How can I optimise it?" he recalls.
Working collaboratively with ChatGPT, he created an automated system using Google Sheets and Google Apps Script:
How the system works:
Union staff enter basic details into a spreadsheet: secretary name, delegate name, dates, and reason for leave
Every few minutes, a Google Apps Script (written with AI assistance) checks if there are any changes to the spreadsheet
When new entries are detected, the script automatically:
Generates a formal PDF letter using a template
Adds the secretary's signature to the document
Sends the letter by email to the employer, the delegate, and the union
Archives the PDF in Google Drive for record-keeping
Updates a database of all leave requests sent
"The whole process now takes about a minute," Stijn explains, "and it happens automatically."
Importantly, Stijn knew nothing about coding before starting this project. "I don't know anything about code writing, but that's why I use AI, of course. It was written with ChatGPT."
The code ChatGPT wrote code is used within Google Workspace, keeping member data secure within ABVV-Metaal’s official systems.
More than 500 union staff who have now been through one of the Unions 21 AI training courses.
What Stijn has done with AI is an example of what our AI training is all about: knowing where and how AI can help union staff focus on what matters most: working with members and workers to build collective strength and improve conditions for all.
Think AI could support you in your work?
We have a few more spaces on our Applying AI for Industrial Officers course remaining. Course starts on 22 October 2025.
Key benefits
Dramatic time savings: What previously took 30 minutes of manual work per letter now takes approximately one minute and happens automatically
Reduced errors: The automated system eliminates transcription errors and ensures consistent formatting
Better record-keeping: Every letter is automatically archived with a timestamp and database entry
Scalability: The system is now being rolled out across all regions of the ABVV-Metaal union in Flanders.
User-friendly interface: The system has recently evolved from a spreadsheet to include a simple form interface that even staff less comfortable with technology can use easily to add details for new requests.
"It saves us a lot of time, Nick. It's crazy," Stijn notes. "Everyone will save time. So it does impact on everyone in the Metaal union, which is a good thing."
Common questions
How difficult was it to implement?
"I started together with AI and [in one] afternoon, I came with a Google spreadsheet," Stijn explains. The initial version was built in a single Saturday afternoon session working with ChatGPT. Later refinements, including adding a user-friendly form interface, took additional time but were accomplished using the same AI-assisted approach.
What skills are needed?
No coding experience is required. "I don't know anything about code writing," Stijn emphasises. However, having a clear vision of what you want to achieve is essential. "I think it's really important that before you do something like that, that you need to have an end goal, that you need to know where you want to go."
What are the main challenges?
The main challenge was making the system user-friendly for all staff members. "Some older colleagues were not used to it," Stijn notes. This led to creating a simplified form interface where users can only make selections from dropdown menus and date pickers, making it "fail-proof."
How do you ensure quality and accuracy?
Because this is a straightforward, repetitive task with consistent parameters (dates, names, addresses), the AI-generated code reliably produces accurate results. "It's most of the time the same parameters," Stijn explains. The consistency of the task means the automated system can be trusted to work correctly without human review of each letter.
What about data protection?
The system operates entirely within Google Workspace, which is GDPR-compliant. Whenever he demonstrates the system at conferences, Stijn uses fictional names from popular culture (Harry Potter, Luke Skywalker) rather than real member data.
Getting started
If you want to try a similar approach:
Identify a clear, repetitive task: Look for processes that follow the same pattern each time, like this union leave letter generation
Define your end goal precisely: Know exactly what you want the system to do before you start
Ask for support from AI: Chat with a generative AI assistant and ask it if it could create code to support you.
Take it step by step: Start with a basic version that works, then refine it based on results and feedback
Consider your users: If colleagues need to use the system, make it as simple and fail-proof as possible
Test thoroughly: Use fictional data initially to ensure the system works correctly
Common pitfalls to avoid:
Starting without a clear vision of what you want to achieve
Making the interface too complex for less tech-confident colleagues
Not building in proper record-keeping and archiving
Forgetting to consider data protection requirements
Looking ahead
Stijn is already planning his next AI automation project: "My next goal is in my free time then to build an AI agent that automatically helps us in generating social media posts or share them from certain people in our union."
The success of this project has inspired colleagues across the union. "I hope I really can inspire people," Stijn reflects. "A lot of people are afraid of AI, me too, in some level, because it goes super, super fast. But the only thing we can do is also embrace it and see how we can use it to optimise our organisation so we have more time for other stuff."
AI Transparency statement
This case study was generated using Claude.ai based on the transcript of a video call between Nick Scott (Centre for Responsible Union AI) and Stijn Merckx (Provincial Secretary, Metal Union of Flanders, Flemish Brabant region).
