It can start with one simple step. Go to ChatGPT, Claude, or any other LLM provider. Type something like: “Teach me AI.” Then use a prompt that: • asks about your current level • understands your goals • and builds a personalized learning plan based on your answers From there, the idea is simple… but not easy. It requires proactivity, effort, and a lot of testing. It means treating the model like a sparring partner. You ask. You try. You build. You break things. And you run everything through your own critical thinking. That’s how you actually learn AI.

I started building my website yesterday, and AI basically became my co-engineer. I made a lot of progress. Drafted the pages, set up the structure, and handled the safety part. And I realised something: I wasn’t building alone. I was building with AI. Any time I got stuck, or didn’t understand how a page works, or broke something, I jumped into ChatGPT / Claude. It felt like having a senior engineer next to me saying: “Try this” “Fix that”. And WordPress on Hostinger has Kodee, their built-in AI assistant, which helped me debug, learn how WordPress works, and understand details about pages, settings, and the whole structure. I still made the decisions. I still wrote the content. I still designed the structure. AI just helped me move faster and learn faster. If you’re building something, use AI as your partner. Not a replacement. A multiplier. The future of making things is not man or machine. It’s the two working in tandem.

And that changes how we learn. I tested Nano Banana Pro a lot more, and I discovered something really powerful. It can generate beautiful, context-rich infographics that actually help you understand things, not just look nice. They can include explainers, structured information, and diagrams based on real-world facts or on content you provide. So I tested it. I made a simple hand sketch on paper about: What AI is. What AI agents are. And what the real practical uses of AI agents are. I gave that sketch to Nano Banana Pro and asked it to turn it into an infographic. The first result already surprised me. It actually tried to read and understand what I drew and completed it with diagrams and structure. After that, I asked it to create a fresh, clean infographic inspired by the design I shared. It generated a version that already looked close to my idea, but in digital form. Then I pushed it one step further and asked it to make it more professional. And this is where it got really interesting. To improve the final result, I even asked the AI itself to help me write a better prompt for creating better infographics. I used that improved prompt, generated a new version, refined it again, and again. And as you can see, each iteration became clearer, more structured, and more professional. So now imagine this at scale. You can take information from documents, books, courses, notes, and research and turn all of that into visual learning material that your brain understands much faster than long text. Would you use AI infographics to learn faster? And what subjects would you want to turn into visuals?

This is the 1-minute version of the 45-minute workshop I delivered yesterday. An AI agent starts with something simple. An LLM. And an LLM is basically a text prediction machine. It guesses the next word based on probabilities. When I type “the quick brown fox” you will see the model pick “jumps” with the highest probability. Same when I type “the president of the US is” you will see it predict “Donald…” and continue from there. (Here is the link to the next-word predictor app: https://lnkd.in/dUiDT2BZ) This is simplified, but this is how it works. One token at a time. On their own, LLMs are static. They do not do anything. They just predict text. An AI agent is what happens when you take this LLM and you attach tools to it. For example, the model can run a web search tool. Or call an API. Or trigger an action. Then it reads the results. Decides what to do next. Maybe calls another tool. And keeps iterating until it solves the task. Agents also include instructions and memory. Short term and long term. So in a nutshell. An AI agent is an LLM combined with tools and reasoning loops. It turns a text predictor into a system that can plan, act, and solve real tasks. #aiagents #ai

You don’t. You learn it by doing projects… and apparently by teaching it. Teaching is one of the best ways to learn, as I found out lately. Right now, I’m building a small AI agents workshop. Everything from the fundamentals of AI, generative AI, and large language models… to the practical stuff like tools, planners, and memory inside agents. And while I’ve been building this material, something interesting happened. Explaining the fundamentals forced me to revisit them. Connect the concepts again. Understand them more deeply. I literally learn better by teaching. And I became a better gen AI engineer in the process. If you’re passionate about a domain, or you want to level up, creating a practical tutorial or a workshop is one of the most effective things you can do. Blend a little theory with hands-on practice. Add a small project. And suddenly you see the subject with new clarity. Because teaching forces you to think deeply. Simplify complex ideas. Explain them clearly. And that process makes you better at the skill itself. And the bonus? You help others grow with you. Everybody levels up.