Techmates at APEK Day: AI in e-commerce without rose-colored glasses

In mid-March, Techmates took part in APEK Day, one of the most important e-commerce events in the Czech Republic. This year, we were not only a partner of the event but also actively contributed to the program.‍

Our Head of Development, Miroslav Kubus, presented a talk focused on the topic:
AI in e-commerce: Successes, pitfalls, and reality without rose-colored glasses.

‍

Why most AI projects fail

Today, almost every company is working with AI in some form. However, there is often a significant gap between enthusiasm and real impact. According to the data presented, around 80% of companies are actively experimenting with AI, but the vast majority of projects never move beyond the pilot phase. Up to 95% remain stuck somewhere between testing and real-world deployment.

The reason is quite simple. AI is often seen as a quick fix for a specific problem. In reality, it affects much more than that. It touches processes, workflows, and decision-making. If it is not approached as a strategic transformation, the results reflect that.

‍

From generating content to taking action

One of the most interesting topics of the presentation was the shift in how AI is being used today and where it is heading.

Most companies currently rely on generative AI, tools that wait for a prompt and then produce text, images, or other outputs. This approach is inherently passive.

Increasingly, we are seeing the rise of so-called agentic AI. Instead of simply responding to instructions, it operates with a goal in mind. It can use tools, evaluate situations, and execute tasks independently. In other words, it does not just provide answers, it actually performs work.

This shift is key if AI is to move beyond experimentation and start delivering real business value.

‍

What AI looks like in real e-commerce

A large part of the presentation focused on a real-world example, expanding an e-commerce business into new markets. At first glance, it might seem like a simple task, translating content. In practice, it quickly becomes much more complex.

A typical e-commerce business deals with tens of thousands of products, hundreds of categories, and a wide range of specific requirements. These include abbreviations, HTML structures, and differences in how products are named across languages.

This is where AI shows its true potential. It is not just about translation itself, but about the entire process around it. In practice, a single tool is often not enough. Better results come from combining multiple models, comparing their outputs, and selecting the best one for each specific case.

Taking it a step further, an agent-based approach can handle the entire workflow. It can detect what needs to be translated, choose the right tool, check the quality of the output, refine it if needed, and store it in the system. It can also monitor SEO quality and flag potential issues along the way.‍

AI without illusions, but with real impact

One of the main goals of our talk was to present AI without unnecessary hype. Not as a magical solution that solves everything on its own, but as a tool that only makes sense when properly embedded into a company’s context.

Experience from real projects is clear. When AI is implemented in isolation, it often fails. When it becomes part of a broader transformation and is connected to real processes, it can bring significant efficiency gains and a strong competitive advantage.

APEK Day was a great opportunity for us to share these insights. It also confirmed that the e-commerce industry is gradually moving from experimentation to real deployment.

If you enjoyed this article, check out our recap from Czech On-line Expo in Prague.