Yup. This is the title I came up with. Well, actually ChatGPT did and that is concerning because I never told it that I love listening to Britney Spears. But I liked it and so this is the title, I simply hope it survives the marketing scrutiny for catchy and SEO friendly titles…

So what’s this article about?

It’s about my thoughts on AI and its impact on ClicData, not just on the platform, or the data teams using it, but also on other areas like marketing, sales, support, service, and product development.

It’s about my impressions after using it personally and professionally for a while, and exploring its broader technical potential within ClicData.

As always, these are just my opinions, nothing more.

They might carry some merit based on my experience and past work, but I’m far from an expert in how these algorithms are actually built. The science, engineering, and development behind them still baffle me more often than not, usually to the point where I abandon whatever thesis, paper, or book I’m reading before the inevitable migraine sets in.

But I do try to understand the theory behind why ChatGPT picked this awesome, Britney Spears related title.

For me, it all started almost 30 years ago.

It’s 1986 And My Atari 520ST Is On Fire

I loved this computer. It was so much better than the green monochromatic Commodore PETs, and the chiclet style keyboards of ZX Spectrum or the bulky, all business like, IBM PC.

The Atari (much like my Commodore 64) gave me graphical freedom, and it somehow opened up a world of possibilities to me and it was mostly because of something called a BBS (Bulletin Board System), a precursor to the internet which my friend Martin and I ran day and night.

At the time, running this BBS was quite a bit of work and I kept writing programs mostly in good old BASIC to automate tasks.

I would write programs to dial into other BBS and pickup mail, transfer game plays, download and upload files, and much more. So much so that I needed a second one to actually run the BBS while the other kept a second modem busy all the time.

But at some point during the night, the bulky power supply of one of the computers started slowly melting the plastic and shortly after caught fire.

The program that, literally caught fire, was named TelmOS and it was an ongoing project of mine to create some sort of intelligent software capable of answering all my questions about the BBS and do certain functions iteratively and without me expressly telling it to do so.

This was what ended our BBS as well. Hard drives were not cheap and with the fire and potentially the fact that the program was running wild, the disk became corrupt and with no possibility of recovery.

TelmOS and our BBS were no longer.

Building the portion of TelmOS that was able to interpret my requests and formulate responses (albeit many of them pre-programmed) was the coolest thing I had done until then.

Seeing GPT as it is today in all its shapes and forms, makes me think of those times and my dream, rather my obsession, of turning a computer into something more.

How to (not) understand Humans

The biggest leap, and likely the most time-consuming, was moving from writing simple programs with thousands (or even millions) of rules for every possible outcome to systems that can actually understand what we type or say.

Natural Language Processing (NLP) was the ability for a program to parse what we say and “understand” the meaning. Human languages are difficult to understand even for those that are highly literate, there is always more than one meaning to anything if you really want it.

To understand speech, the complexity intensifies with not only dialects and slang entering the mix but also accents and pronunciations.

Today, you can say [tomayto] or [tohmato] or even [tomatoe] and GPT will understand it, probably faster than you typing it and all this without writing an IF statement for each variation or using SOUNDEX or other more archaic ways.

And yes, the fact that GPT can then provide lengthy answers to the question or have a conversation with a human is impressive (more on this later), the fact that we are able to program (teach?) a machine to understand us humans, to me remains the most fascinating thing about this process.

“Telmo, do you really think they understand us?”, you may ask.

My answer is “yes.” In as much as a dog understand the word “sit”. It is able to understand the commands and interpreting them correctly–although I had a dog once that 3 times out of 5 would actually walk away instead of sitting.

And that is all there is to understanding isn’t it? Its not difficult. If you understand something and are able to react to it, is that intelligence?

Not sure. And personally I am not sure if that is even the point.

We can start arguing over whether machines can think or if they are conscious, morality, right from wrong, or if they can harm us. If that is the case then I am afraid there are many non-intelligent human beings as well. And did I mention that my dog once bit me because I took his tennis ball straight from his mouth? Who’s the less intelligent one?

Irrespective of philosophers, as far as I am concerned, we do not have Artificial intelligence, but we do have a high level of Machine Interaction coupled with the ability to learn, basically store questions and responses to determine which ones are most asked and accepted (reinforcement).

At the surface this seems an oversimplification of what is nonetheless quite impressive coding, and with the increase in performance of the hardware behind these computations (mainly around GPUs), we are now seeing a near real-time interaction with the machine at a lower cost.

As with any other major industrial and technical milestone, the phenomenon known as GPT has had an impact on the way we work.

As we slowly discover its usefulness in accomplishing certain tasks, we are now using it at home, at school and at work.

So what are some of those impacts and where do we get the most benefit of this new tech?

At School

As with many things in life, I didn’t even get the chance to introduce my kids to the world of GPT, they were already using it, and mostly for the wrong reasons: outsource their homework!

Of course that presents a few risks like not retaining anything from the repetitive task of applying what you learned during the day. Repetition at an early age (and again at a later age) is critical and basically having someone else do the homework assigned to you is not the way to learn.

Assuming that GPT provides the right answer, and typically that is the case, the process of doing homework has boiled down to taking a photo of the exercise book and asking GPT to provide the answers.

On the positive side, some times, and mostly when doing less repetitive tasks such as projects or reviewing their lessons, they would prompt GPT to explain the topic or problem to them. This proved to be quite good for a few reasons:

• Neutral Tutor: By removing the parent from the equation it also removes the pressure that kids feel, especially in their teens, from having being taught or told. This is the time they try to set it out on their own and having the parents explain concepts and helping them is simply not “cool”. It also relieves the pressure on the parents on either having to figure out things that have been long forgotten and probably are not taught the same way anyway.
• Clear Steps: By nature, GPT provides step by step and can adapt to the way it explains things by simplifying terms and even providing analogies, without sounding like an adult. In fact, even when I am helping my kids with certain subjects, I use GPT to help me put it in a clearer and better way.
• Always Available: There will be no longer a “I’m cooking dinner” or “busy with my own meetings” excuse.
• Adaptable: The curriculum from one school system to another is not always the same and GPT adapts to the region and language. I found this extremely helpful as it would disserve my kids if I was to help them as I was taught when I was their age.

Overall, if you keep an eye on the “homework replacement” factor, GPT is a very positive tool for my kids academics.

At Work

Marketing

GPT has been made part of the marketing job description, and that’s pretty clear to anyone reading blogs, scrolling social media, or opening emails.

Most content scream AI and we quickly notice GPT-generated videos, blogs or images because they’re bland without creativity, and repetitive of other content.

Nonetheless, there are certain features such as orchestrating the thoughts of the marketing team and focusing them on specific topics as well as content that could potentially be used.

Once again, GPT handles the mundane tasks that take up much of our marketing team’s time, helping them focus on improving content and structuring the message more effectively.

Another thing it does is help reshape the content once it’s written, by correcting grammar and making it more accessible to the reader. I find that’s where GPT brings the most value in marketing.

For small and medium-sized businesses, GPT’s use in image and video is still quite limited. Similar to written content, it does well with standard tasks like editing, refining, or improving what already exists. But it still falls short when it comes to creating content that feels fresh, interesting, and visually engaging.

Creativity is potentially the last barrier that machine learning algorithm needs to break–also the one thing that machines can’t learn from existing content.

Sales

I believe that sales can and should use GPT in most activities, especially for prospect qualification as GPT is connected to social and business networks. The capacity for it to provide a summary of the person or people that a sales person is about to talk to is a huge time saver.

Additionally, technology, such as transcript and summarization of calls and meetings, can condense hours of work on a weekly basis in to minutes.

Software Development

Our largest team at ClicData is made up of software developers and engineers and interestingly not all of them jumped on the GPT band wagon right away.

I believe they mistrusted this technology and used it seldomly and only for basic things such as getting them started on a function, a new module or a new construct, or just to get the bare bones structure down. Again, it’s about saving time on repetitive (and boring) tasks.

Today most of them go beyond that and use it to better their code and identify potential bugs. Much in the same way that Marketing uses it to improve an existing base of work, they use it to make their code more robust and clean.

They also use it to create test scenarios, prepare sample data sets, and simulate payloads for testing or developing against some sort of a base structure.

They also use it to get better understanding of certain topics such as authentication methods, new frameworks, and knowledge base type of questions.

Support

Our support team deals with many interesting problems on the daily basis. Many of those are repetitive in nature and as you know, this is where GPT excels. The ability to turn a 3-minute support ticket response into a 30-second reply adds up significantly over the lifetime of a support team member.

Additionally, GPT can also provide a learning opportunity to all support team members. In our case some support tickets are more challenging than others, involve multiple technology, different databases and different systems. By quickly prompting GPT about these systems and databases, the support team can advanced much more rapidly in identifying the potential issues.

GPT also comes in very handy when some customers require additional information such as corrections to their database queries or styling of their dashboards.

Currently, we are working on training GPT on ClicData and its help system in the hopes that not only our support team to take advantage of it but also our customers. We feel this will be a game changer for many of them.

Services

Similar to Support, our services team is embracing GPT in the areas of advancing certain technical tasks such as SQL queries, dashboard formulas, and data transformation.

Much like the development team they also use it to create sample data sets so that they don’t work with customer or production data. Data anonymization takes a long time especially when doing it across multiple related datasets.

Our services teams are data experts (data engineers, scientists, analysts) but they are not subject matter experts in every single industry and business process. Here, GPT can also assist greatly by providing the team with terminology and insight into each industry and its KPIs.

Finance

Finance benefits from GPT in many ways but one that I take advantage is understanding accounting rules and regulations across the different companies we operate in. I also use it to validate my Business Plan ensuring that the plan has no mistakes or bad calculations and is coherent.

Most recently we adopted Pennylane, an accounting platform, that has AI baked in it which greatly facilitates many tasks, especially of matching bank transactions with invoices.

At Home

More recently, I’ve been thinking about adopting GPT more seriously at home. We sometimes use it instead of Google to clarify something or just to joke around with its voice, but I’ve been thinking about how to make it more useful, rather than just treating it like a fun gadget.

I have been thinking about how old Alexa and Siri are and how they are so behind the times, and I find myself repeating turn off living room lights, 3 or 4 times before they understand what I want (memories of my dog come alive). And the thing is, these are commands that I have been repeating for years, every day sometimes more than once a day and yet, these devices do not learn and do not evolve.

With the number of home devices constantly probing, listening, measuring and controlling heating, windows, lights, security and more, I think it’s about time GPT reaches the home automation arena.

What AI Taught Me, Other Than Fire Safety

In just 3 years, AI became an integral part of how we work, learn, and interact with technology, but its role goes beyond just automation—it’s changing the way we think about intelligence and machine interaction.

From my early experiments with automation on my Atari 520ST to today’s use of GPT in marketing, sales, software development, support, and finance, it’s been a fascinating journey. AI is great at handling repetitive tasks, structuring information, and assisting in complex problem-solving, but it’s not without its limitations.

At home, in education, and even in business, AI’s benefits are undeniable, but so are the risks—whether it’s students using it to avoid learning or companies trying to replace human judgment with algorithms. The key is understanding how to use AI as a tool for enhancement rather than a replacement for creativity and critical thinking.

As we continue to integrate AI into more aspects of life, from automating tasks to improving decision-making, its potential is exciting—but only if we use it wisely.

At best I hope this blog gave you some ideas and maybe even had you explore the impact of AI in your company, your family or your life. At worst, you now know my dog didn’t listen to me and I set my Atari on fire.