From chatbots to autonomous agents: what has changed

Just two years ago, AI automation mainly meant text generators and "copilots" that assisted humans with writing or programming. Today the situation is different. Companies are deploying agentic AI — systems that don't wait for a prompt but recognize on their own what needs to be done and carry it out.

According to Gartner, by 2028 up to 80% of enterprise workflows will use AI agents. Automation is moving from simple conditional rules to independent real-time decision-making.

What does this mean in practice? An invoice arrives by email, AI reads it, extracts the data, enters it into the accounting system, and sends a payment confirmation to the client — all without a single human click. And that's exactly what the tools we're introducing today are all about.

n8n: Open-source automation with an AI brain

n8n (pronounced "n-eight-n") is a fair-code workflow automation platform that has experienced rocket growth over the past year. Unlike commercial competitors, it offers self-hosting capability — your data stays on your servers, which is crucial for companies handling sensitive data or subject to GDPR.

n8n connects over 400 apps and services — from Google Workspace through Slack, Notion to databases like PostgreSQL. In 2026, AI-native nodes were added, enabling:

• Classify and analyze incoming emails using LLMs
• Generate summaries from internal documents
• Create agentic workflows where AI decides the next step based on context

Price: The self-hosted version is free (Community Edition). Cloud plans start at €20 per month for small teams. For Czech companies, it's important that n8n supports custom LLM endpoints — you can connect even European models running locally, without sending data to the US.

Make: Visual automation that even understands Czech

Make (formerly Integromat) is a European platform headquartered in Prague, making it one of the most interesting players for Czech companies. It offers a visual scenario editor where you assemble individual automation steps like puzzle pieces — no coding required.

In 2026, Make integrated AI modules that can:

• Convert voice recordings to text and directly create tasks from them
• Generate personalized emails based on customer behavior in CRM
• Analyze sentiment in reviews and automatically escalate negative feedback

Price: The free tier offers 1,000 operations per month. Paid plans from $9 per month (10,000 operations). Business plans with advanced AI modules from $34 per month. Czech localization is partial — the interface is in English, but support and documentation are also available in Czech.

Zapier: Automation veteran with AI Central

Zapier remains the largest player in the market — it supports over 7,000 integrations and its strength lies in the breadth of its ecosystem. The 2026 novelty is Zapier Central, where an AI agent acts as the "brain" of the entire automation and decides on its own which action to trigger.

Zapier's AI agent can:

• Monitor the inbox and automatically respond to routine inquiries
• Create content drafts based on data from Google Analytics and social networks
• Manage the CRM pipeline and reassign leads by score on its own

Price: The free tier offers 100 tasks per month. Paid plans from $19.99 per month (750 tasks). For full AI agent capabilities, you need at least the Professional plan at $49 per month. Czech is not natively supported, but AI models can handle Czech text.

Microsoft Power Automate: When AI sits right inside Office

For companies using Microsoft 365, Power Automate is the logical choice. With Copilot integration in 2026, you can describe automation in natural language — just write "when an email with an invoice arrives, save the attachment to the Finance folder and enter the due date into Excel" and Power Automate creates the scenario on its own.

Key advantages:

• Deep integration with Teams, Outlook, Excel, SharePoint, and Dynamics 365
• Support for RPA (robotic process automation) for working with legacy applications
• AI Builder for extracting data from documents without model training

Price: From $15 per user per month. Premium licenses with RPA from $40. Complete Czech localization — interface, help, and documentation.

Agent platforms: When one agent isn't enough

For more demanding scenarios, multi-agent platforms like CrewAI or Relevance AI are gaining ground. These allow you to create a team of specialized AI agents that pass tasks between each other — similar to having a virtual marketer, salesperson, and analyst who communicate with each other.

CrewAI is an open-source framework for agent orchestration. Each agent has a defined role, goal, and tools. A typical scenario: a browser agent finds potential client contacts, a copywriter agent prepares a personalized email, and a planner agent schedules a meeting in the calendar.

Relevance AI offers a visual interface for building agent teams without code. Their specialty is sales and support automation — agents connected to CRM, e-shop, and customer support in a single dashboard.

CrewAI price: Open-source free, cloud plans from $15 per month. Relevance AI: Free tier for individuals, team plans from $29 per month.

What does this mean for Czech companies?

The Czech market is specific — it's dominated by small and medium-sized enterprises that don't have the budget for enterprise solutions like UiPath or full AI transformations. That's precisely why tools like n8n (free for self-hosting) and Make (with development in Prague) are the most interesting for Czech companies.

The question of data sovereignty is also important. European regulation (EU AI Act, GDPR) pushes companies to keep track of where their data goes. Self-hosted n8n or European clouds (Make runs on AWS in the EU) are therefore a sensible choice.

At the same time, automation does not equal layoffs. According to the McKinsey Global Institute, by 2030 Europe will see a transformation of job roles rather than massive job losses. Routine tasks will disappear — freeing up capacity for creative work, strategic decision-making, and client care. Exactly as the "experience economy" trend describes: when AI automates routine, value shifts to human contact and experience.

How to choose the right tool?

Deciding factors when choosing:

• Team's technical proficiency: n8n requires basic technical thinking, Make is more accessible for non-technical users
• Integrations: Check whether the tool supports your existing applications. Zapier leads in connector breadth
• Budget: n8n Community is free, Make offers the best price/performance ratio in basic plans
• GDPR and data sovereignty: Self-hosted n8n or Make (European company) are safer choices
• AI capabilities: For advanced agent scenarios, bet on CrewAI or Relevance AI