Automate the boring stuff

Make Windows your batch

This is a small article about computer automatization. Why do all the repetetive tasks yourself when you can have your computer do them for you? I think this is something that everyone who is working on a computer can benefit from. And today this is easier then ever, you do not even need that much coding experience anymore, just an LLM that can help you get going.

Everybody seems to be talking about artificial intelligence snagging away our jobs, yet nobody seems to know when this will happen exactly. This makes it even harder for us as more or less average employees and workers of regular office jobs to know where we could improve ourselves or our abilities to cope with those machines. But from how things stand today, I think we can agree that the machines are catching up.

Why not let the computer do the boring stuff for you?

The most impressive feats the llms can do from my point of view are creative in nature. But why would anyone in their right mind try to take that from us humans? I for my part really love the creative aspects of my work and instead try to hand off more and more of the repetitive and boring stuff to the machine. Everytime I see someone or myself not making even use of basic automation properly I sometimes think to myself: “No wonder that we are doomed for replacement.”

It is 2018 2026. Computers have been around for a while now and are so ubiquitous that it is hard to understand why so few people are inclined to use more of the “deeper” functionality of these machines. I do not speak of full on programming your own software, but about thinking systemic. Get into scripting. Start automating the boring stuff, because hey, noone can do a large amount of repititive tasks better, more efficient and less error-prone than one of those pesky machines that one day will eat our jobs. The biggest advantage of starting with repetitive tasks is that they are annoying and the time saved will compound so fast that it will be a real win for you. On top of that they can often be performed as simple folder actions or as scripts running in larger software applications so that you can get going without a lot of overhead.

Why scripting is such an easy win

There are many ways of getting started with scripting. MacOs has the Automator and Folder Actions that make it really easy. So easy in fact that you don’t have any excuse to not automate repetitive tasks. Windows on the other hand does not have any of those niceties and has always preferred to show you everything at once all the time, it even brings three different Terminals for scripting and I never know which one to use (but luckily it is getting easier and with WSL you can just use the regular old terminal for most things). And Linux. But when you use that, you can stop reading this article, because I can probably bring nothing new to you.

Automate the boring stuff

The computer is made for chaining together ad infinitum any number of repetitive steps you throw at it. And since it does not (yet) have it’s own agenda it is quite content to sort through files, rename heaps of images or aggregate the content from dozens or hundreds of PDFs into one big file all while de-duplicating the content.

“Well”, you say, “renaming images is not that big of a deal” You may be right. But what if you are not talking about ten or twenty images, but instead about hundreds or thousands or hundreds of thousands? Takes a little longer now. And what about when you not only have all those images in one big folder but in multiple folders? Will you manually drill down all of those folders to find the images to be renamed? Well, okay. But one year from now, when it is time to move the next years worth of content into the archive, then you do it all over again, manually. Using the exact same renaming scheme? And how about deriving a part of the image name from the actual content inside of the image? This example is by no means an extreme one. It is just one of the many repeating tasks that crop up in a small or medium sized office working in the field of design, where you often have to manage a large amount of lots of different file types to produce a delivery for you client.

And what if I say you could write a few different lines of code for each of the different tasks, save it as a folder action and from now on looking for files, grouping files by type or by metadata, renaming them and moving them to where you actually need them becomes a breeze.

Or you can chain all of those scripts together to form a workflow that can be used by your colleagues for consistency. Or that can even be automated itself, like say it listens to a server to runs automatically whenever a client has uploaded new files.

Barrier of entry

The example above sounds very compelling yet the vast majority of computer workers still do all those things manually by hand when they instead should focus on automating the repetitive tasks to free up time for more creative or personal work that they are there for in the first place.

On a second note, I have to admit that scripting is not as easy as I may have made it sound. The Apple Automator is the closest to being easy I came across while scripting on any computer and makes it really easy to get started. But even the Automator has no easy to follow along tutorial. And while I use a combination of Powershell, Batch and Python to automate MacOs and Windows tasks (mostly Python nowadays) I still fumble around regulary. And if it wheren’t for sites like “Stackoverflow”, “Poweruser” and “Reddit” I probably would have only automated not even halve of the stuff I have.

AI to the rescue

Scripting is one of the things that LLM really excel at. Just tell your Chatbot of Choice what you want to do. It does not really matter which one to choose, they are almost interchangeably useable when you are just getting started and a lot of them have generous free tiers to get your feet wet. When trying out new ideas I really like to use the Model Fusion by Open Router https://openrouter.ai/fusion where you can enter a prompt and then have it be sent to a selection of different LLMs and their answers are then fused into on output. This is really great for complex ideas, but overkill if you are just getting started.

AI as a teacher and a guide

The best thing about the LLMs is that you can just tell them to teach you through it. And instead of reading through the same generic tutorial that does not translate to your setup or system and which you then have to arduously debug while scouring Stackoverflow, you can get really specific, say “teach me through it” and the results are fantastic. And if you do get stuck on a really hard problem along the way you can just screenshot the problem and ask the LLM to help you fix it.

What a time to be alive.

In resumen: get scripting

Over the years I have met many people that are very reluctant to pick up scripting as a tool for their workflow, even if most of them can see the potential it has to improve it. But when the prospect of shaving hours and days of repetitive work off your To-Do list and freeing up that precious time for other things is not enticing enough, I don’t really know what is going to make you start to pick up coding.

But instead of a change of tools I see the problem more in a missing explanation of the already available tools, how they work and what they can unlock. By spending just a few hours learning the basic concepts of scripting and how computer programs work, you can start to create very simple scripts that in a few days or weeks will recoup more time for you than you have spent writing them. That should be enough to whet your appetite. Not everyone wants to explore that on it’s own.

This is one of those tasks where an employer should actively invest in it’s employees, because ultimately it is their workflow that would benefit the most. To me, all that talk about artificial intelligence replacing our jobs still comes a little too early. We should stick to the machines we already have on or under our desk and get started using good old fashioned scripting.

Don’t burn them trees, when Python suffices.