What is agentic AI and why the department is deploying it

Agentic artificial intelligence (agentic AI) refers to systems that, unlike traditional chatbots, can independently plan, make decisions, and perform multi-step tasks without constant human supervision. While a regular chatbot answers a question and waits for the next instruction, agentic AI schedules its own workflow, gathers the necessary information, and delivers the result.

Acting Chief Technology Officer of the Department of Commerce Matheus Passos explained during a GovExec webinar on April 24, 2026, that the department sees agentic AI as a tool for workforce empowerment, not its replacement. "We believe agentic AI will be a great tool for the Department of Commerce to empower our workforce. Not to replace anyone, but to enhance them. It will make them better, smarter, and make their jobs easier," Passos said, according to MeriTalk.

Nine pilot projects — from acquisitions to project management

The department is currently testing nine specific use cases of agentic AI. All of them focus on structured, repeatable workflows where automation can operate within clearly defined boundaries. The pilot projects include:

• Acquisition and procurement support — according to Passos, this is an ideal entry point because the processes are highly structured and rule-driven
• Job description creation — automation of routine administrative tasks in recruitment
• Project management — milestone tracking, team coordination, and reporting
• Customer relationship management (CRM) — more efficient contact and communication management

Deployment takes the form of a three-month onboarding program that begins with broad familiarization with the technology and gradually transitions to practical agent development directly in the hands of employees. It also includes enterprise-wide training designed to dispel fears and explain how the technology will actually be used — Passos noted that anyone who fears being replaced by AI simply won't use it, even if it means remaining less productive.

"One Commerce" — agentic AI as a unifying tool

The Department of Commerce has a strongly federated structure — it encompasses agencies with very different missions, such as the Census Bureau, the Patent and Trademark Office (USPTO), or the National Institute of Standards and Technology (NIST). Each has its own agenda and data systems, which has historically led to organizational silos.

The "One Commerce" initiative seeks to break down these barriers. According to Passos, agentic AI is meant to play the role of a great equalizer: "Agentic AI is a great equalizer — it can look across the organization and start coordinating efforts in areas where people have completely different missions," he explained. In other words, an AI agent can connect data and processes across agencies that previously rarely communicated with each other.

Security first — trust as a condition for success

A key condition for deploying agentic AI remains security and data protection. The department tests agents in controlled environments and laboratories, especially with sensitive data related to economic indicators and national security.

Passos put it unequivocally: "If a manager or anyone in the workforce believes their data is not secure, they won't touch it. They'd rather be less productive." Employee trust is thus, in his view, a fundamental prerequisite without which the entire initiative would fail. "Our number one priority is often to keep up with vendors and ensure that security vulnerabilities are addressed," he added.

European contrast: regulation versus voluntary standards

The U.S. Department of Commerce's approach offers an interesting comparison with Europe. While the U.S. relies on voluntary standards (such as the NIST AI Risk Management Framework) and a market-based approach, the European Union has opted for comprehensive regulation — the EU AI Act, which comes into full effect in August 2026, establishes legally binding obligations for developers and operators of AI systems, including agentic solutions.

At the same time, the European Commission actively supports the development of its own AI solutions for public administration — its library guide on agentic AI was updated in May 2026, and EC policy speaks of "accelerating the deployment of European generative AI solutions in public administration." The World Economic Forum issued a special framework in April 2026 for assessing government readiness for agentic AI.

In the Czech Republic, no systemic initiative for agentic AI in public administration exists yet. While the U.S. Department of Commerce is already testing concrete agents for procurement and project management, Czech authorities are only just discovering basic chatbot capabilities. However, in the context of the EU AI Act, even Czech institutions will need to define their approach to deploying more autonomous AI systems in the coming years.

This is not the endgame

Passos also emphasized that agentic AI is not the destination, but just another phase of technological development. "Agentic AI is not the endgame — the next thing will come. We give you this tool in your hands while also thinking about how to improve it and what comes next," he said. For the Department of Commerce, this is therefore not a one-off project, but a permanent cultural and technological transformation aimed at making civil servants more productive professionals — while also preparing the ground for the next generation of AI tools.