Artificial intelligence has become an integral part of the IT work environment. From software and hardware specialists to ordinary users. It offers tools that can automate routines, find errors in code or create complex reports from data. The key is to know its limits and risks. Use it as a powerful assistant, not as a smart new colleague.
Generative models such as ChatGPT, Gemini or Copilot can process text, images or video and create content based on the calculations. Therefore, it is important to validate answers, not work blindly with the results and use critical thinking. AI is a tool, it does not replace humans.
Practical use in companies and risks
Protecting personal and company data is key. Inserting sensitive data into public AI systems, be it internal documents, source code or client data, can lead to leaks. While some models claim that they no longer handle or store it, can we trust that?
Generative models are designed to always answer, regardless of whether the question is accurate, nonsensical, or even tends to end the conversation.
Therefore, it is important to check answers carefully and be aware that repeating questions frequently or tackling the same problem from different angles can lead to stuckness. That’s when the AI starts to hallucinate. Why is this happening?
In layman’s terms: Artificial intelligence works by turning words, sentences or images into vectors. That is, long series of numbers capturing the meaning and relationships between concepts. During training, it learns from vast amounts of data and creates an internal “map” that helps it guess what word or information should come next. When generating answers, she always calculates the probabilities of possible continuations and selects the most likely one. So at the core, it’s just complex math and statistics, not conscious thought.
From chatbots to agents
Only a few years ago, these models were rather unknown, not only in Slovakia. The first systems were able to recognise images or translate text. Their reliability was limited. Today they are a practical tool for everyday work. They produce text, images, music and videos. Their emergence has ushered in a new era of digital productivity. From simple chatbots, they are AI agents that can plan, execute tasks and learn back from the results. Their capabilities are growing exponentially. Modern models can now process tens of thousands of words of context at once, work with multimodal inputs (text, voice, image, video), and perform actions directly in the digital environment, not just provide answers.
How we use AI in practice
Also at GAMO we test, develop and integrate applications that can independently process emails, create tables, write a piece of code and pass the finished output to another application for processing or direct use. In practice, this means that processes that took minutes to tens of minutes a few years ago can now be handled in seconds to minutes. At the same time, we use multiple agents for different tasks. To give you a better idea, we will mention a few minor ones.
Directives Agent: targets all internal directives documents stored and regularly updated in SharePoint. It does not matter if the file is in Word, PDF or Excel format. When a question is asked, it responds immediately, retrieving and reading the necessary data from the files and formulating it into a direct, understandable answer.
Instructions agent: works on a similar principle to the Directives, but also has information from the company intranet. In the context window in the intranet/SharePoint application you enter a question: What are the current regulations for this year for installed applications on the company phone and how do I install them?
At that point, both agents activate, look for specific information and merge it into one coherent answer. If necessary, they will also supplement it with references to the documents from which they have drawn. This process will make it much easier for a new employee to navigate company policies or assist with a company phone change. There really are many possibilities.
The main idea is to get rid of the manual routine and bring the end user the result in the required time. The applications we use on a daily basis can summarize an email, process notes from a meeting or prepare a response. However, what adds the most value is the ability to pair a question asked by a human with an internal knowledge library and specific data so that the result is an accurate and immediate answer. Such a process can replace dozens of minutes of work by one or more colleagues.
The biggest transformation of work
The evolution of generative AI in recent years shows that this is not just a technological advance, but a fundamental transformation in the way we work. In 2022 came ChatGPT and with it the feeling that “AI has finally arrived in the office”. Just four months later came the significantly more powerful GPT-4. Companies realised that it was no longer just another IT tool, but a technology that could write, summarise, program, design and communicate in a way where the line between tool and ‘colleague’ blurred. From this point on, a global race was on to see who could integrate generative AI into internal processes faster and more efficiently. From testing to pilot projects to massive deployment.
According to current McKinsey analysis (2024-2025), companies can already automate up to 60 percent of administrative tasks and increase the productivity of knowledge workers by 20 to 40 percent. In customer support, AI reduces the volume of calls handled by humans by up to half, in marketing it creates first versions of campaigns, in sales it acts as a virtual salesperson, and in IT it acts as a co-programmer. This turns AI into a super assistant that takes on most of the prep work, while the human plays the role of editor, strategist and guarantor of the outcome.
Yet the speed of development never ceases to surprise. Anthropic managed to increase the capacity of the Claude AI model from 9,000 tokens to 100 thousand tokens of text in two months, which translates to about 75,000 words per minute. An incredible leap in such a short time. The year 2025 was also riding on this wave: Meta introduced the multimodal models Llama 4 Scout and Maverick, Google launched Gemini 2.5 Pro with a million context window and a “thinking” mode. The trend is clear. We have moved from simple copilots to autonomous agents capable of planning, decision making and action in real time.
The challenges and responsibilities of working with AI
With increasing adoption come new risks. Generative AI can not only assist, but also hallucinate, spread inaccuracies, unwittingly violate copyright, reinforce biases or create manipulative content. Companies therefore need to set clear rules, protect privacy and sensitive data, strengthen cybersecurity and ensure quality control. Humans must remain in the decision-making process for critical tasks.
AI is changing the way we work. It has gone from being a helper to digital agents that can take over part of the work process. The current period represents the ‘deployment phase’. Companies are linking AI to internal systems, building their own agents, and setting the stage for the corporate AI assistant to become as common as email. Those who master productivity, people and ethics all at once will get a head start. The others will just catch up, as with every other technological revolution.
