Artificial intelligence is moving beyond assisting businesses to running them. Researchers describe the model as a “zero-person company,” where autonomous agents carry out all operational work, such as planning tasks, using digital tools, and execute business processes with no direct supervision.
The concept is approaching technical feasibility. In late 2025, researchers from KPMG and the University of Amsterdam ran an experiment to see whether a group of AI agents could operate a small online business. The agents managed a digital shop selling personalized AI-generated art. The exercise showed that software agents increasingly can perform tasks that once required human employees.
From a tax perspective, this raises an important question: What happens when a company has no people performing work and no human managers taking operational decisions?
Before addressing that question, it’s important to clarify what a zero-person company actually means.
Human Anchor Needed
Despite the name, a zero-person company can’t exist entirely without people. A company is a legal construct, and corporate law requires a human link somewhere in the structure.
No jurisdiction allows a company to incorporate without a traceable connection to a natural person. A director, founder, or ultimate beneficial owner must exist somewhere in the ownership chain. That person provides the legal accountability for the entity. Operational decisions may be delegated to software, but legal responsibility remains tied to a natural or legal person who acts as the founder or director.
Although value-added tax rules were developed with human-run businesses in mind, the system is flexible enough to apply to companies operated by software. The fact that technology performs all operational tasks and makes business decisions doesn’t change the legal analysis.
A simple analogy helps explain why. A company that sells drinks through vending machines may have no employees present at the point of sale. Yet the company is considered to be running a business there. The machine is simply a tool used to conduct the activity.
AI agents play a similar role. They perform tasks on behalf of the entity that deploys them. Even when those tasks include planning, sales or customer interaction, the company itself still carries out the economic activity. The presence or absence of employees doesn’t alter that basic principle.
Place of Establishment
The more difficult question is where a zero-person company is established for VAT purposes. The country of establishment matters because it determines whether supplies are treated as domestic or cross-border and whether VAT must be charged on services supplied to business customers.
Under Article 10 of Council Implementing Regulation (EU) No. 282/2011, the place where a business is established is the place where the functions of its central administration are carried out. Several indicators help identify that location, including the registered office of the business, where management meetings take place, and where key decisions are made about general management. Priority should go to the location where essential management decisions are taken.
For a traditional company, these indicators usually point to the same place. Executives work in a specific location, and strategic decisions are made there.
A zero-person company complicates this analysis. If AI systems execute operational decisions, and the company has no employees performing management functions, it becomes difficult to identify a physical place where central administration occurs. AI agents typically operate through distributed cloud infrastructure without a meaningful geographic location.
In such cases, the registered office may become the most visible indicator of establishment. However, EU case law shows that the registered office alone isn’t always decisive. In Planzer, the Court of Justice of the European Union emphasized that a registered office without real management activity may not be enough to determine where the business is established for VAT purposes.
This creates an interesting tension for zero-person companies. If software is handling operational decision-making and the registered office lacks operational substance, tax authorities may look beyond the formal structure. They may examine where the human founders or directors are located to determine where central administration effectively takes place.
Fixed Establishment Unlikely
Another issue is whether a zero-person company could create fixed establishments in other countries where its systems operate.
Under Article 11 of Council Implementing Regulation (EU) No. 282/2011, a fixed establishment requires a sufficient degree of permanence and a suitable structure in terms of human and technical resources that allows the business to receive or supply services.
The requirement for human resources is central. The CJEU repeatedly has confirmed that technical infrastructure alone generally isn’t enough to create a fixed establishment if the taxable person doesn’t have its own personnel in that jurisdiction.
This interpretation has clear implications for zero-person companies. The presence of cloud infrastructure or automated systems in several countries is unlikely to create fixed establishments on its own. However, the analysis may change if one of the human founders is based in another country. In that situation, tax authorities could argue that the human presence contributes to a structure capable of supporting a fixed establishment.
Compliance Without Employees
Many aspects of VAT compliance can already be automated. Software can generate invoices, maintain transaction records, and file VAT returns. But the tax systems still assume that a human counterpart exists somewhere in the structure.
Tax authorities issue information requests, conduct audits, and ask questions that require explanations about the underlying transactions. Even when software prepares the data, someone ultimately must respond and take responsibility for the company’s tax obligations.
For that reason, a completely human-free company remains more theoretical than practical. AI systems may perform most operational tasks, but the legal framework of VAT still depends on identifiable human actors who can be held accountable.
This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law, Bloomberg Tax, and Bloomberg Government, or its owners.
Author Information
Aleksandra Bal is global tax technology lead at Stripe and a frequent contributor to tax publications and industry conferences.
Write for Us: Author Guidelines
To contact the editors responsible for this story:
Learn more about Bloomberg Tax or Log In to keep reading:
See Breaking News in Context
From research to software to news, find what you need to stay ahead.
Already a subscriber?
Log in to keep reading or access research tools and resources.