Today, April 23, 2026, represents one of the most significant milestones in integrating artificial intelligence into everyday work. Microsoft officially announced that agentic capabilities of its Copilot are now fully available for Microsoft 365 users. What does this mean for us? The difference between a "chatbot" that writes an email for you and an "agent" that prepares an analysis, compiles a report, and subsequently creates a presentation from it, is profound.
From Chatting to Autonomous Action: What Is "Agentic AI"?
To understand the significance of this update, we need to explain the difference between a traditional LLM (large language model) and an agentic system. While a common model like GPT-4 or Claude 3.5 responds to your instructions within a single conversation, an agent possesses the ability of planning and reasoning.
An agentic workflow means that when you enter a task like: "Analyze sales for the last quarter in Excel, find the biggest trends, and prepare a summary in Word and a presentation in PowerPoint from it," Copilot will not just generate text. It breaks down the task into several subtasks, visits Excel, performs calculations, extracts data, moves to Word, assembles the document structure, and finally creates visual materials in PowerPoint. This process runs with minimal human intervention, while the AI itself decides which tools and steps to use to achieve the result.
This approach aligns with the trend we see in models like OpenAI o1, which focus on deeper reasoning, but Microsoft has a huge advantage in the practical implementation of these capabilities directly into the work environment thanks to its integration into the Office suite.
Practical Use in Key Applications
Microsoft Word: From Writing to Editing
In Word, Copilot will no longer be just a "text enhancer." Thanks to agentic capabilities, it can work with extensive documents as an autonomous editor. For example, it can take dozens of pages of notes from various sources, unify them into a single style, check factual accuracy against external data, and create a structured document with a table of contents and citations. For the Czech user, this means huge time savings when creating contracts, technical documentation, or academic papers.
Microsoft Excel: Analysis Without Formulas
Excel has always been a barrier for many users due to the complexity of formulas. The agentic Copilot in Excel can perform complex data analysis based on natural language. Instead of writing complex macros or formulas, you can say: "Find anomalies in our transportation costs over the last three months and create a graphical interpretation for management." The agent itself identifies relevant columns, performs statistical tests, and prepares a visualization.
Microsoft PowerPoint: Visualization Automation
Creating presentations is often the most demanding part of work. Copilot in PowerPoint can now take a complex document from Word and not only create slides from it, but also design a logical presentation structure that matches the presenter's goals. It can automatically select appropriate visual elements and ensure the presentation has a professional look without the need for manual arrangement of elements.
Comparison with Competition: Who Leads the Battle for Office Efficiency?
In the field of AI assistants, a fierce battle is currently underway among three main players:
- Microsoft Copilot: It has the biggest advantage in the ecosystem. Integration directly into applications that companies around the world (including the Czech Republic) use is unbeatable. Agentic capabilities within Office are currently the closest to real office work automation.
- Google Gemini (Google Workspace): Google offers very strong integration into Google Docs, Sheets, and Slides. Gemini is excellent in speed and is a natural choice for users of the Google ecosystem, but Microsoft still holds a stronger position in the enterprise segment (large companies).
- OpenAI (ChatGPT/GPT Agents): OpenAI offers the most advanced models for reasoning itself, but their "agents" are often isolated within a chat interface and thus lack the deep integration into local documents and applications that Microsoft has.
Price and Availability for the Czech Market
For Czech users, it is important to know that Microsoft 365 Copilot is also available in Czech localization, which allows interaction with the agent in our native language.
Pricing Policy:
- Copilot Pro (for individuals): Approximately 20 USD / month (roughly 470 CZK when converted). This plan is intended for home use and personal productivity.
- Copilot for Microsoft 365 (for businesses): Approximately 30 USD per user / month. This plan is intended for the business segment and includes advanced security features and enterprise-level data protection.
It is necessary to emphasize that an active Microsoft 365 subscription (Personal, Family, or Business) is required to use these features.
Security and EU Regulation
For European companies, compliance with the EU AI Act is a key topic. In its announcements about agentic capabilities, Microsoft emphasizes that Copilot operates within "Enterprise Grade" security. This means that your data, which the agent uses to perform tasks, is not used to train public OpenAI models. For Czech companies subject to GDPR, this data isolation is absolutely fundamental to maintaining the confidentiality of trade secrets.
Do I have to write special commands for every function in Excel?
No, thanks to the agentic mode, you can use natural language. You can enter a goal (e.g., "Analyze this data") and Copilot itself will decide which steps and tools it needs to use to deliver the result.
Is Copilot fully functional in Czech even for complex tasks?
Yes, Microsoft is gradually expanding Czech language support for all Copilot capabilities. Although some very specific technical terms may require English, basic agentic tasks in Word, Excel, and PowerPoint work in Czech.
Will Microsoft ensure that my sensitive company data does not leak to the public internet?
Yes. Microsoft guarantees that data used within Copilot for Microsoft 365 remains within your organization and is not used to train base AI models, which is in line with European data protection standards.