Back in 2022, I watched an episode of WAN Show by Linus Tech Tips about the release of the OpenAI ChatGPT AI bot, it is was our first exposure to AI changing our world. It’s now 2026 and this is my thoughts on AI and how I see it.
Now, I want to be specific here and say that I have 100% authored this post, and zero AI has been used.
WAN Show AI
This WAN Show episode was the time that really opened my eyes to what ChatGPT was capable of. I had heard about its weeks leading up to this, and didn’t really take notice or had some initial skeptism about whether people that were hyping it up were being honest with what it could do.
Following this episode, I signed up and gave it a go, was immediately blown away with what it was able to do to with what seemed like very little effort. I was getting it to write stories for my kids, write PowerShell scripts, answer questions, explain things. It was the future, right there on my phone, in my hand.
Being Amazed
Chat-based AI was nothing we hadn’t seen or experienced before and now the world was amazed and becoming obsessed. Tasks that would take hours or days would now take minutes. Tasks that would take teams of people to complete could now be done in moments while you went to the Cafe for a coffee. And businesses were quick to want it for their business.
Into the Office
For my work, we were looking for ways to both implement AI into our workflows as well as to have answers for our clients. They wanted it quick, but we knew that without the proper setup they weren’t goint to get out what the sales pitches promised. This was particular accurate when it came to the release of Microsoft Copilot.
Those businesses in the Microsoft 365 world, Microsoft 365 Copilot was sold as the safe option for businesses as it ensured that business data will remain safe and guarded from public Large Language Models (LLMs) as well as within the business, guarding employees from asking about information that they didn’t have privilege to. Business was now able to generate reports in minutes from simple notes, and turn that report into a professional PowerPoint presentation, again reducing human effort from hours to minutes.
Our workplace had a range of AIs being used by people from ChatGPT, Google’s Gemini and the consumer-version of Microsoft Copilot. Since our business was ISO 27001 certified, this was a no-no, so we moved and mandated Copilot being the AI of choice, whether it was the best or not it meant we were now using what we were selling and educating our clients on. And then it all evolved from just doing word documents and emails, do building entire software development solutions. What would have taken a team of software developers, could now be done by someone that had zero software development experience or even understanding thanks to Anthropic‘s Claude AI.
Claude-ing it Up!
The last 6-8 months, I’ve been playing with different AI models and what they are good at and which one gives better responses for the work and prompts I provide. And this might be obvious but this is where this is at:
- ChatGPT
General AI work, Q&A, image generation, some code - Microsoft Copliot
Great for the workplace using the Microsoft suite and its ability to take work data and make something from it. - Claude
The only AI to use for software development.
The results that Claude is able to produce when it comes to software development brings me back to that moment when I original tried out ChatGPT in 2022. Holy crap could it do amazing things. We are now producing our own in-house apps since there’s no off-the-shelf apps that work with our workflows 100% and now each team can have apps that do the work. All done without a single software developer on staff. This is vibecoding and it’s now a fully acceptable practice.
I had (and some days still have) reservations about videcoding in our business being a technical guy. I had a lot of questions:
- Who’s maintain this?
- Where are the servers going to be hosted (especially long-term)?
- What’s the security look like?
Especially when these apps were using APIs to connect our main ticketing/CRM system.
Still all early stages of this process and I was asking big question about something that was just a playground. Seeing what we COULD achieve, and what COULD work for us that other software couldn’t do, or what would be costly monthly subscriptions for a 90% product. Claude allowed us to do create a solution that was ours, and that was OK. But it also worked, and worked well. No programming, no training, just a prompt.
Where it’s going and I dont like it
My whole IT career up to this point has been do the hard yards of training, working in the trenches, learning how to do the processes the hard way before leveling up and doing the better work. Spending hours some times doing the difficult work. Doing the research yourself, because no-one else had the answers and everyone was looking to you to solve the problem. The DNA of being in IT. With AI, that’s over.
The engineers entering or already starting out in the Level 1 or Level 2 Help Desk don’t have to face these difficulties. The Entry-level software engineer doesn’t need to study the books and the language manuals to understand how to write the software and how to debug it when it all goes wrong. Hard hours are no longer a requirement. It’s now accept that if you’re no using AI as part of your workflow, then you’re doing it wrong and your wasting everyone’s time.
For me personally, being “old man yells at cloud” type means that it’s difficult for me to accept that a new generation of techs and engineers will have this easy route to diagnosing their problems. Hell, my first boss used to have Microsoft TechEd on CD that he had installed on his laptop to look up Error Codes since being online 24/7 wasn’t a thing. So, getting instant answers to questions, in a chatbot on your EXACT issue with internet access is mind-blowing in its own right.
I don’t like it because people are going to quickly lose skills that took us so long to develop. AI was built off the backs of experts that refined their craft over decades. In the span of four years, all of that has been absorbed by the machine and spits it out without credit or acknowledgemt of those that did the grind. This is applicable for software development, writing, music, art and more. Everyone is affected and AI relies on human to input creative so it can learn how to be human, and those entering the workforce dont need to understand HOW it works, just need to know how to ask the correct AI prompts. That’s the need job skill.
AI Here to Stay
AI is here to stay and is very much part of our lives, even if planned Datacenters are being cancelled. This situation will continue to evolve for better or for worst. Seasoned techs like me need to either adapt or will be left behind like when computers started taking over jobs of those that would calculate math and draw spreadsheets with paper and pen.
How well can you direct the AI will be the new baseline. Decades in the future, there will be someone that will be looking back at ChatGPT 3.5 and laughing about how useless it was and wonder how we ever achieved anything with it in comparison.
This article is a snapshot of this state of changes in the industry and how I personally value that hard yards, knowing how to do something manually before automating it. And how I we need to protect our ability to be creative before we lose it to the system in favour of quick results for profit.
What are your thoughts on the state of AI? Please leave a comment below and check out some of my Microsoft articles.