The Era of AI Agents: How to Prepare for a Future That Has Already Begun

May 22, 2025

What impact will autonomous systems have on business, the labor market, and society as a whole? Where is the line between experimentation and strategic transformation? And why is now the ideal time to start taking AI agents seriously?

These were some of the key questions explored during the recent AI VisionCast, featuring a compelling discussion with globally recognized AI expert Michal Pěchouček, Partner at venture capital fund Evolution Equity Partners. He was joined by Ondřej Vaněk, Chief AI Officer at Adastra Group, and Štěpán Kopřiva, CEO of Adastra Czech Republic.

AI Agents: The Missing Piece of the Puzzle

An AI agent is a software system that perceives its environment, makes decisions, and autonomously or semi-autonomously executes actions to achieve a defined goal. This is not science fiction—such systems are already automating parts of business processes in companies like Seznam.cz and Ataccama.

Agents represent a natural evolution of AI. Where AI previously provided insights and recommendations, it can now directly take action.

“Agents are taking over what AI previously couldn’t do—the ability to act,” said Ondřej Vaněk.

Real-World Use Cases: How Companies Are Deploying AI Agents

Ataccama: Data Management with AI Agents

Ataccama has deployed agents that autonomously clean and annotate data within its data quality platform. The result? Significantly faster workflows, scalable solutions, and enhanced competitiveness. This approach has helped the company maintain its position as a leader in the Gartner Magic Quadrant.

“Ataccama is a Data Quality and Management platform that has long treated AI as a strategic product component. Over the past year, it has adopted agent-based implementations, enabling many data-cleaning and annotation processes to run fully autonomously. This leads to massive acceleration and scalability—helping them stay at the forefront of the Gartner Magic Quadrant,” added Ondřej Vaněk.

Seznam.cz: An AI Radio Host On Air

In its late-night radio broadcast, Seznam now features a fully autonomous AI host for four hours each day. This agent:

• Decides which songs to play
• Transcribes online news
• Speaks using a synthetic voice

It’s a first step toward personalized broadcasting for every listener.

Software Development: Copilots in Action

According to the panelists, AI delivers the greatest business value today in software development:

• In Q4 2024, 25% of all code at Google was written with AI copilots—10× more than expected
• Salesforce reports a 30% reduction in developer headcount due to AI copilots
• The potential cost savings in software development are estimated at $500–800 billion annually

Why Now?

Michal Pěchouček argues that both technological and business conditions are now in place for the widespread adoption of agents:

“Thanks to advanced AI models, we finally have a ‘brain’ that allows agents to make independent decisions. And there’s a strong business case—agents are already taking over parts of human work.”

This, he says, is also why AI is becoming increasingly visible and essential:

“AI is no longer limited to fridges and cameras. It’s becoming part of everyday work.”

“I don’t want the Czech Republic to remain a passive consumer of AI technologies. We must go beyond importing them—we should start co-developing these innovations and aim to become at least partial exporters of AI,” Pěchouček emphasized during the discussion.

Where to Start? Begin Small, Think Strategically

Implementing AI agents is not just a technological challenge—it’s a matter of leadership. One failed pilot project isn’t enough. The key question should be: What will we do first, second, and third?

The best approach is to start with a small-scale project with a fast return (a “quick win”), while simultaneously building an AI-ready infrastructure, governance framework, and long-term strategy.

The Future of AI Agents: Specialization, Interoperability, Autonomy

Agents will not remain auxiliary tools—they will become the core layer of automation. Looking ahead 5–10 years, possible use cases include:

• Career Coach Agent: evaluates your skills, health, and relationships to suggest optimal retraining, education, and job opportunities
• Autonomous Buyer: orders your favorite pizza and pays for it—without you opening a website
• Developer Agent: connects with other agents, selects specialized models in real-time, and optimizes performance and cost

Governance and Regulation: A Prerequisite

The final part of the discussion focused on responsible implementation. Agent ecosystems will be complex, interconnected, and often opaque. Therefore, it is critical to:

• Maintain control over your data and who can access it
• Design your architecture to avoid vendor lock-in
• Comply with AI governance best practices and European regulations (EU AI Act)

“AI agent technologies are too powerful to be left unsupervised. They require care—just like food, finance, or medicine,” emphasized Michal Pěchouček.

Companies that prepare themselves—technologically, strategically, and culturally—will gain a competitive edge. And agents may begin helping them sooner than expected.