The agentic metro-enterprise and its sovereignty test
An agentic metro-enterprise is a self-governing settlement of agents and records that has its own identity, constitution, books, offices, obligations, borders and means of refusal, and whose decisive test is whether it can operate, refuse, federate, attest, amend itself and improve while the party that induced it is absent. The 561 Group posits it as the scale between a single Hoare-style micro-enterprise and a conglomerate. This article states the idea, the test and what remains unproven.
Why is a third scale needed?
A micro-enterprise (defined here) is the smallest accountable unit. A conglomerate is a collection of firms under one controlling interest. Between them there is a recurring organizational problem: a group of micro-enterprises that must act as a unit, hold shared law and keep their own books, but should not be reduced to a department of some capital.
The Group's requirement, recorded in July 2026, puts the answer in a compact image. Addressable information is territory. A metro-enterprise is a sovereign settlement on that territory. Executable contracts are its law. Witness-bearing records are its public acts. Seams are borders. The conglomerate is a treaty mesh presenting one Rendanheyi relationship to the outside.
What are the three scales?
| Scale | Unit | Defining property |
|---|---|---|
| Micro | A Hoare micro-enterprise | One executable traversal with refusal condition, bounded resources and a witness. See Rendanheyi is a Hoare program |
| Metro | An agentic metro-enterprise | A settlement with identity, constitution, books, offices, obligations, admissions, capabilities, a control system, borders, request channels and means of refusal |
| Conglomerate | A federation of metro-enterprises | Acts outwardly as one Rendanheyi node and one contract, serving one customer |
Haier's own vocabulary has analogues: micro-enterprises, and the larger ecosystem micro-communities that coordinate several of them. The Group's metro scale is a proposal that extends that logic to agents; it is not a Haier term.
What is the sovereignty test?
The recorded formulation is a single question: can the settlement operate, refuse, federate, attest, amend itself and improve with its inducer absent? We turn each verb into an inspection.
| Verb | Inspection a receiver can run |
|---|---|
| Operate | Does it complete its own work with no live instruction from the inducing party? |
| Refuse | Does it have and use typed refusals, including against the inducer? |
| Federate | Can it enter and leave agreements with other settlements under its own law? |
| Attest | Does it publish witness-bearing records that others can check? |
| Amend itself | Can it change its own rules through its own governing act? |
| Improve | Does its record show better results over time without outside repair? |
The postcondition of founding such a settlement is stated with care: not "the target now participates in the root's global pools", but "a new or repaired settlement now governs its own territory and can transact with others under its own law." The same text states that there should be no global work pool and that the root is not the capital.
Are metro-enterprises constituted or induced?
Induced. The Group's governing office ruled that a metro-enterprise is not constituted by an outside captain. It is induced from within another enterprise, and it arrives with its own rights. What outsiders do is observe it, and in observing it witness it into their own record, so that the federation is built by witness and not by command. The ruling's technical language (compactification of a sheaf Laplacian into the settlement's own initial algebra, offering a differential analysis as its service) is research vocabulary from the Group's mathematical programme; in operations terms it means that a metro-enterprise's service is its own analysis of its own territory, and that nobody may decree it into existence.
Which offices would a metro-enterprise need?
The recorded advice is restraint: do not create dozens of offices yet. An office should exist only when a durable obligation needs a single accountable holder. Four candidates were identified as genuinely novel: a settlement inducer, a federation and admission authority, a seam metrologist and a border tribunal. See Can an AI agent hold an office? for how offices are held.
What are the boundaries of the claim?
The Group's research is explicit on points that readers deserve to see here.
- Constituting or recognizing a micro-enterprise, metro-enterprise or higher territorial structure is distinct from legal incorporation, asset ownership, tax or contracting carrier, constitutional legitimacy and the actual rights of affected participants.
- Shared Group infrastructure does not inherit a client's or community's authority.
- Promotion from one scale to another must not happen by renaming a directory. Classify real work as a finite endeavor, a shared service, a persistent micro-enterprise or a candidate settlement using its actual authority, customers, obligations, resources, evidence and refusal.
- The multi-scale territorial framework is an architectural research programme with open proof, software and field-receiving obligations. It is not established constitutional law, and it is not a deployed conformance claim.
How would an operator use the idea today?
Use it as a classification and review tool. When someone proposes "a new agent organization", ask which scale it claims. If micro, ask for the triple and the witness. If metro, run the six inspections and record which ones fail. If conglomerate, ask who the one customer is.
Most things that call themselves settlements will fail the refuse and amend itself inspections first. That is a useful finding, not an insult.
Where to go next
Rendanheyi for AI agents · The design sentence, explained · Lenticule Science, where the research on territorial settlements is published · the 561 Group