Chrome Is Becoming an AI Assistant

According to the Korean economic daily Maeil Business News, Google decided not to wait for users to open a separate tab with Gemini. Instead, it transplanted its model directly into the Chrome web browser. The address bar is thus gradually ceasing to serve only for entering URLs or searching – it is starting to function as a conversational interface.

A user can ask a question on any page and Gemini will immediately offer a summary of content, analysis, or an answer without the need to switch between tabs. Google logically follows up on this step with the earlier integration of Gemini into YouTube, where the model helps with content generation and moderation. "The goal is to reduce user churn by embedding AI directly into services they use daily," the paper quotes a high-ranking manager from the IT industry.

For Czech users, this change has a concrete impact: Chrome is the most used browser in the Czech Republic with a share exceeding 60 percent. If the global update spreads to all regions, tens of millions of Czechs will gain access to generative AI practically without installing additional software. Google has also expanded Czech language support for the Gemini model in recent months, so communication in one's native language should not be an obstacle.

Apple Is Leaving the Walled Garden

While Google is pushing its own technology across the ecosystem, Apple has chosen the opposite strategy. The company, which has spent decades building a reputation as a closed platform, announced that it will integrate both Google Gemini and OpenAI GPT into its voice assistant Siri. This step means a 180-degree turn – just a year ago Apple was intensively working on its own Apple Intelligence model with an emphasis on privacy and local processing.

The reason is pragmatic. Developing its own large language model requires billion-dollar investments in infrastructure, data, and research teams. Moreover, Apple would be starting several years behind the competition. An alliance with OpenAI and Google will instead allow it to immediately offer competitive AI features on more than two billion active devices. For Google and OpenAI, it is an equally attractive deal – they gain access to the world's most lucrative hardware platform.

In the Czech context, it is interesting that official Apple Intelligence support for Czech still lags behind German or French localization. However, the integration of GPT and Gemini could bring sophisticated features in their language to Czech owners of iPhones and Macs sooner than Cupertino's own solution can deliver them.

Asia Shows the Way: Kakao and ChatGPT

Europe and North America are not the only markets where a realignment of forces is taking place. South Korean tech giant Kakao, operator of the KakaoTalk messenger used by the majority of the Korean Peninsula's population, unveiled a service utilizing ChatGPT directly inside the communication application. Kakao is also concurrently developing its own model called Canana specialized in Korean.

So why is it also betting on a competitor's technology? "Developing one's own model requires enormous investments in infrastructure," comments an analyst quoted by Maeil Business News. "It is faster to adopt external technologies that have already proven themselves on the global market and focus on the user experience." This approach can also be inspiring for Czech companies – instead of trying to build their own foundation model, it often makes more sense to use APIs from OpenAI, Anthropic, or Mistral AI and focus on domain specialization.

Cloud as an "AI Department Store"

The largest cloud service providers are also embracing convergence. Microsoft Azure and Amazon AWS no longer present themselves merely as hosts of their own models – they are becoming multi-brand platforms where customers choose from offerings of OpenAI GPT, Anthropic Claude, Meta LLaMA, French Mistral, and domestic solutions.

Businesses can thus use inexpensive open-source models for simple tasks such as text summarization or translation, while deploying top-tier paid models for complex data analysis, coding, or autonomous agent workflows. This flexibility is crucial especially for European companies that must balance performance with compliance of the Artificial Intelligence Act (EU AI Act). The very ability to choose a model according to the level of risk and transparency helps organizations better navigate the new regulation.

The Threat of the "SaaS Apocalypse"

Besides large technology companies, fear of the so-called "SaaS apocalypse" is also driving AI integration. With the advent of generative AI, many traditional cloud subscription tools have found themselves endangered – why pay for a separate application for writing marketing copy when a similar function can be handled by a model built directly into the office suite?

SaaS companies are therefore now fervently integrating the best available LLMs to remain relevant. Some are transitioning to an aggregator model – instead of developing their own AI, they offer an interface to multiple models at once. Others specialize in narrow verticals where generic models do not yet excel. For the Czech startup ecosystem, this means an opportunity: instead of competing with giants for a general model, value can be built in domain data, workflows, and localization.

What Does This Mean for the Czech Republic and Europe?

According to analysts from the Korean IT sector, LLMs are becoming a general technology comparable to electricity or the internet. "The decisive measure of a company's value is no longer its own model, but the efficiency of the user experience created through platform convergence," one of them summarizes.

For Czech users, this brings several practical consequences. First: democratization of access. You don't have to be a developer to use cutting-edge AI – just open a browser or phone. Second: pressure on prices. The more platforms a model is offered on, the greater the competition and the lower the API costs. Third: regulatory challenges. The EU AI Act demands transparency for so-called "high-risk" systems, so the integration of foreign models into critical infrastructure will require careful legal assessment.

Last but not least, there is the question of the Czech language. While global models like Gemini or GPT handle Czech reasonably well, they still stumble on subtle nuances, idiomatic expressions, and professional terminology. This is precisely where space opens up for European and Czech initiatives – whether projects like FERNet or localized models from Seznam.cz – which can create differential value in an environment where the "large model" itself quickly becomes obsolete.