For a long time, we were accustomed to the era of "chatting" with AI. We type a query, GPT or Claude generates a response, and we either accept it or correct it. This scenario, however, is changing dramatically. According to reports from CIO.com, the LIV Golf league has decided to leverage agentic AI to deepen fan engagement. These are not simple chatbots, but intelligent systems capable of interacting with data in real time and delivering personalized experiences that were previously impossible.
What exactly is Agentic AI? The difference between a model and an agent
To understand why this shift is important, we must clarify the difference between a standard large language model (LLM) and an agentic system. While a model like GPT-4 or Claude 3.5 Sonnet functions as an extremely smart "brain" that can predict the next words and understand context, Agentic AI is this brain connected to "hands" and the ability to plan.
Agentic AI has three key characteristics:
- Reasoning: The ability to break down a complex goal (e.g., "Find me the best moments of today's tournament based on my favorite player") into individual logical steps.
- Tool Use: The agent can decide on its own that it needs to use a calculator, search data in an API, or open a statistical table.
- Autonomy: The agent can correct itself within a given task if it encounters an error, without having to wait for further instructions from a human.
In the context of LIV Golf, this means a fan no longer has to manually browse statistics. The agent can tell them in real time: "Given the current wind and ball position, your favorite has a 65% chance of a successful putt."
Technology background: Salesforce and Adobe leading the wave
This trend is not just about sports. The big players in software have long prepared the infrastructure for these systems to be safe and reliable. Salesforce recently introduced the Agentforce Testing Center. It is a tool designed for managing the AI agent lifecycle. For businesses, this means they can test their autonomous agents in safe "sandbox" environments using synthetic data to avoid hallucinations and errors in production.
Adobe is moving in a similar direction. At the Adobe Summit 2026, it presented tools capable of creating entire marketing campaigns in minutes using AI agents. This is no longer just about image generation, but about systems that can plan a campaign, select a target audience, and then execute it.
Comparison of technological approaches
| Feature | Standard LLM (e.g. GPT-4) | Agentic AI (e.g. Agentforce) |
|---|---|---|
| Main function | Text/code generation | Task and workflow execution |
| Interaction | Reactive (waits for prompt) | Proactive (plans steps) |
| Tools | Text context only | APIs, web, databases, applications |
Practical impact: What does it mean for Czech businesses and users?
You might ask: "What do I care about golf?" The answer is simple: the technology powering LIV Golf is the same as that which will be used by Czech e-shops, banks, or public administration.
For the Czech market, this has several levels of impact:
- Customer support: Forget frustrating chatbots that just repeat FAQs. Agentic AI in an e-shop can actually process a complaint, check stock levels, and create a new order.
- Efficiency for small businesses: For Czech small and medium enterprises (SMEs), Agentic AI can act as a "virtual employee" that handles administration, reducing operating costs.
- Regulation and safety: Here comes the key point for Europe. Given the strict EU AI Act, any autonomous agent that decides on a user's fate (e.g., in banking) must be transparent and auditable. Companies in the Czech Republic will have to invest in agent monitoring tools, similar to what Salesforce is doing with its Testing Center.
Price availability and market
If you are considering implementing these technologies, you must count on them being enterprise solutions. While basic models like ChatGPT are available within a Free tier or for 20 USD/month, agentic systems like Salesforce Agentforce or Adobe Experience Cloud are priced as part of complex corporate subscriptions, often running into thousands of dollars per month depending on data volume and number of agents. For the average Czech user, however, these capabilities will gradually appear within regular applications (e.g., in Google Workspace or Microsoft 365) as part of their standard subscriptions.
In conclusion
The shift from generative AI to agentic AI is one of the most significant shifts in software in the last decade. The LIV Golf case shows that it is a tool for building a relationship with the user, not just a technological novelty. For Czech businesses, it is a signal that they must prepare not only for how to use AI, but above all for how to manage, test, and control it to be in compliance with European safety standards.
Can agentic AI systems be dangerous if they make a mistake?
Yes, which is why Agentic Lifecycle Management is key. Without testing in isolated environments (sandboxes), an agent can cause an error in a database or an inappropriate interaction with a customer in a real environment. That is why companies like Salesforce are investing in monitoring and testing tools.
Is Agentic AI available in Czech?
The models themselves (language foundations) like GPT-4 or Claude are very capable in Czech. The agentic systems themselves (e.g., within Salesforce or Adobe) depend on how a company configures them. Most large enterprise platforms support Czech, but their full implementation requires local data and rule setup.
What is the difference between Agentic AI and a standard automated process (RPA)?
RPA (Robotic Process Automation) is "blindly" programmed: if a button on a website changes, the RPA fails. Agentic AI has the ability to reason — if the interface changes, the agent will try to understand the new situation and find a way to achieve the goal, making it a much more flexible system.