From BHFF Scholar to Speaker: Vladimir Mijić on Working with AI Agents - What They Give and What They Take Away

At BHAAAS 2026, BHFF alumnus and AI/ML engineer Vladimir Mijić delivered a thought provoking presentation titled “Working with AI Agents: What They Give and What They Take Away,” opening an important discussion about how AI is reshaping software development and the way we work with technology.

In a very short period of time, AI tools have evolved from experiments into everyday habits. While interfaces may not appear drastically different, what happens behind the scenes is fundamentally changing how engineers think, build systems, and make decisions. AI models are moving beyond simply generating answers toward taking actions and autonomously executing tasks.

During the presentation, Vladimir explored the shift from traditional LLMs to AI agents capable of analyzing tasks, gathering context, using tools, executing actions, and iterating on results. A special focus was placed on where AI agents deliver the most value, particularly in tasks with clearly defined goals and verifiable outcomes.

One of the key questions raised throughout the session was:
Are AI tools actually making us more productive?

While AI can significantly accelerate implementation and code generation, software engineering has never been only about writing code. Understanding the problem, evaluating trade offs, making sound decisions, and maintaining systems over time remain core engineering responsibilities. Faster development does not automatically mean better systems.

The presentation also highlighted the growing challenge of “cognitive debt,” where systems evolve faster than our understanding of them. As the amount of generated code increases, so does the need for deep understanding of architecture, domain knowledge, and the long term consequences of technical decisions.

Another important takeaway was that AI should be used as a tool for learning and support, not as a shortcut that replaces thinking. Good practices include asking for explanations, comparing approaches, and reviewing personal solutions, while blindly copying generated answers can slow long term skill development.

The conclusion of the presentation was clear:
The value of engineers is shifting from the speed of writing code toward the quality of the decisions they make.

AI will continue reducing the cost of implementation, but knowledge, critical thinking, and understanding complex systems remain irreplaceable.

As highlighted at the end of the presentation:
“Knowledge still pays off.”