In an era of rapid artificial intelligence advancement, even the most seasoned experts can be surprised by the pace of innovation. The Build with Gemini conference, organized by Google, wasn’t just about presenting the breakthrough Gemini 2.5 Pro model—it was primarily about hands-on workshops and actionable knowledge that empower developers to better leverage AI in their daily work.
Aleksandra Szałajko, an experienced ML engineer at Comarch, shares her insights from the event. She talks about the innovations in Gemini 2.5 Pro that impressed her most, how AI is transforming developers’ daily routines, and what new opportunities are emerging for teams thanks to Google’s latest tools. We invite you to read this inspiring conversation about the future of technology and the role of AI leaders in business.
Ola, what made you decide to attend the Build with Gemini conference?
From a professional standpoint, the decision was quite straightforward. The Google brand and the quality of its products alone ensure that we’re dealing with top-tier technology. The Gemini 2.5 Pro model, the main focus of the conference, is currently considered one of the best in the world. As an ML engineer, I’m always eager to expand my knowledge of cutting-edge AI tools, and this conference was a perfect opportunity to see Gemini in action and evaluate its potential for use in our projects.
Have you participated in Google-organized events before, especially those targeted at engineers?
Yes, we’ve had a working relationship with Google for a while. Two years ago, our entire AI team was invited to the newly opened Google office in Kraków. We were introduced to their latest cloud solutions and products. Around the same time, we also gained access to a special Google training platform, and I actively took part in the courses offered there. It was a great opportunity to deepen our understanding of the AI tools available through Google Workspace.
What were your first impressions? Did the conference meet your expectations in terms of organization and content?
The event was excellently organized—rather than general marketing slogans, we received a solid dose of technical knowledge, code snippets, and practical workshops. It was clear that the presenters were true experts with a passion for sharing complex ideas in a clear, accessible way. Most importantly, the conference was filled with practical information and examples that we plan to use in our projects. Google clearly focused on developers—putting tools in our hands and showing us how to use them. It was solid engineering knowledge, not just theory.
Which of the presented innovations caught your attention the most? What do you consider a breakthrough feature in Gemini?
I was especially intrigued by the feature that enables chatbots to express emotions. The fact that conversations with AI can increasingly resemble interactions with real people—not just machines—has enormous potential in terms of business user interfaces. But the most groundbreaking feature, without a doubt, was Veo 3—a tool that enables high-quality video generation from text descriptions. The ability to describe a scene in words and then watch it come to life on screen is truly stunning.
Comarch Team at the Build with Gemini conference (Aleksandra – second from the left).
How do you evaluate the direction in which Google is developing AI, particularly in the context of Gemini?
I think the pace of Gemini's development is astonishing. Breakthrough features are appearing one after another—things we couldn’t even imagine AI models being capable of just a short time ago. I’m talking about things like realistic video generation or recognizing and expressing emotions. Google is clearly aiming to create AI that’s not only analytical but also more human and creative.
With this level of innovation, do you see a possibility to implement Gemini or similar models in your own projects? Which areas do you find most promising?
Yes, we already have some plans—but specifically with the Gemma model. The key advantage for us is that it’s open-source, so we can freely use its functionalities in our commercial solutions. Gemma offers several crucial features that will allow us to expand and enhance our projects.
To better understand Gemini's capabilities, which workshops and sessions did you attend? What stood out and proved useful in your work?
I participated in three workshops focused on language model evaluation and agent-based system design. I also attended several lectures—mainly on features we're already developing in our projects, as well as those showcasing the latest capabilities of Gemini 2.5. The most valuable for me was the workshop on LLM (Large Language Model) evaluation. I’ve been deeply involved in this topic while developing a similar feature for our ChatERP project. The session provided me with additional insights and introduced new tools that will help significantly improve the accuracy and efficiency of our language model evaluation module.
Has attending these sessions changed your perspective on AI? Did anything surprise or inspire you?
What surprised me the most was the sheer speed of Gemini’s development—and how quickly these groundbreaking technologies are becoming practical tools. Not long ago, features like multimodality or AI-based video generation seemed like a distant future. Today, we see that multimodal capabilities are becoming standard. The vision of using a phone as a personal assistant—one that understands context, processes images, and performs complex tasks—is no longer futuristic. It’s happening right now.
There was also a demo zone at the conference—did you get to test any of the tools yourself? What were your impressions?
Absolutely—and that was one of the most exciting parts of the conference! In addition to lectures and workshops, there was a special demo area with booths run by Google employees. They showcased their products, and you could test them freely and ask questions. What impressed me most was an interactive game. I had the chance to play it and see how the graphics were personalized in real-time based on user preferences. Even the gameplay varied each time, as the AI algorithms analyzed player moves and adjusted future stages accordingly. It was a fantastic way to experience how creatively AI can be used.
In your daily development work, how would you describe the role of AI, especially Gemini? How does it impact your efficiency and work quality?
AI tools like Gemini are becoming essential partners in a developer’s everyday workflow. As intelligent copilots, they significantly speed up code writing and bug fixing by automating repetitive tasks and suggesting relevant improvements. In the design phase, features like deep research act as advanced analysts—synthesizing data from multiple sources to propose the best architectural patterns and technologies. As a result, these tools streamline the entire development process and allow me to focus on more creative and strategic challenges.
After such an intensive event, what are you taking back to your work—knowledge, ideas, contacts? How do you plan to use them?
Above all, I’m taking back a wealth of practical knowledge that I can directly apply to our projects—especially regarding language model evaluation and implementing safety mechanisms. The workshops and lectures sparked several new ideas for evolving our solutions, which we’ll definitely start testing soon.
Lastly—if you could ask Gemini one question that only AI could answer, what would it be and why?
That’s a tough one! As an engineer working on systems that optimize business operations, I’d ask: “Considering all publicly available data on global supply chains, financial flows, and interdependencies between economies, what single optimization process would generate the greatest positive impact on the global economy?” I’d be very curious to know whether the answer would be something obvious—or something we as humans have completely overlooked.