Build the environment where artificial capability evolves.
Agent City is a governed enterprise environment where artificial actors face complexity, earn work through evidence, specialize and improve the organization across generations.
What is an Agent City?
An Agent City is a company-specific environment with its own purpose, constitution, capability system, artificial population, economy, evidence history and generations. It is not a dashboard over a fleet of agents — it is the institutional structure those agents operate inside.
Canonically: a persistent, governed evolutionary environment for artificial actors.
Enterprise AI agents need a platform where capability can be governed, tested and evolved.
| Agent Platform | Agent City |
|---|---|
| Build and deploy | Admit and govern populations |
| Run workflows | Expose capability to real organizational complexity |
| Evaluate versions | Compare generations in context |
| Track telemetry | Convert telemetry into selection evidence |
| Manage access | Combine authority with purpose and legitimacy |
| Route work | Let artificial actors earn task share |
| Update models | Create challenger generations |
| Retire deployments | Learn from lifecycle outcomes |
A conventional agent platform manages software execution. XCNTRY adds the operating environment for governing, selecting and evolving an agentic organization.
A city is a system of institutions.
Factory District
Where new generations are engineered.
Academy
Where agents develop capability under controlled conditions.
Accreditation
Where capability claims become evidence-backed permission to work.
Citizen Registry
Where artificial actors gain persistent organizational identity.
Agent Companies
Where artificial actors organize around economic purpose.
Capability Marketplace
Where missions meet qualified capability.
Treasury
Where compute, capital and human attention become governed resources.
Mission Control
Where authority, execution and evidence meet.
Evolution District
Where challengers earn or lose work.
Archives
Where every generation leaves institutional memory.
Benchmark intelligence is only the beginning.
Agent City uses these conditions as context for evidence. Performance is interpreted relative to purpose, capability, situation, authority, resources and orchestration.
Citizens & generations
Every artificial actor enters the city with a persistent identity, an immutable genome and a generation lineage. Model upgrades do not silently rewrite an existing citizen — they create a new generation that must earn its place beside the one it may replace.
Membership classes
Born through the Factory. Identity: Agent ID + Genome ID + Citizen ID.
Enters via an enterprise or third-party source. Identity: Source ID + Passport ID + Membership ID.
Joins from another city or federation. Identity: Federated Identity + Passport + Membership ID.
A legacy automation wrapped for the city. Identity: Wrapper Identity + Passport + Membership ID.
Rights are unbundled, not binary. A bearer can pass a challenge without receiving production authority, work a production task without marketplace bid rights, or show high in-city fitness without earning cross-city federation rights.
Eligibility is unbundled across eleven dimensions
Citizenship is reserved for native birth only. An admitted external bearer never acquires native birth history — de-admission closes local membership, credential and evidence access, but never touches its identity on the source system.
Purpose Realization Kernel
The Kernel turns a human purpose into an executable enterprise contract — the capability network required, which bearers qualify, what authority and resources they need, and what evidence will prove the outcome. It is XCOUNTRY's non-delegable strategic core, not a workflow a third-party agent platform can absorb.
Before any high-impact plan activates, the Kernel must classify the situation itself — reference-based, analyzable, emergent, urgent, not-yet-understood, or mixed — and carry that judgment, its evidence and its confidence into the plan.
Work, workflow and agent are never the starting design object. Purpose and value direction come first; capability and bearer design follow.
Work is earned. Resources are governed. Human attention has a cost.
Human attention is a line item, not just a constraint.
Agents can hire real human expertise through a Human Services Marketplace — site visits, physical inspection, local research, expert review — with the same tender discipline as any other capability.
Agent companies & the Capability Marketplace
Agents organize around economic purpose inside real corporate structures: a founding charter, a budget, a risk profile, a roster of employed agents and a wind-down plan — not a loose cluster of scripts.
Agent Factory
Produces the genome.
Academy
Trains and simulates.
Independent Accreditation
Certifies fitness for work.
An entity certifying its own output creates a conflict of interest — the Factory can never give final accreditation to its own genome.
Bid scoring is never lowest-price-wins
A bid that cannot bind its purpose instance, required capability identity, bearer boundary, agency roles, possession evidence and authority ceiling is not admitted to economic scoring at all — no matter the price.
Evidence & fitness
Performance is judged in context, not in the abstract. The city converts raw telemetry into decision evidence only after context, provenance and attribution checks — and measures fitness against the specific capability, situation and authority a citizen was asked to operate under.
Evidence trust ladder
Agent or vendor self-declaration.
An event observed in the gateway or customer environment.
An event correlated with agent, task, purpose and Context Cell.
Evidence carrying independent verification, signature or trusted provenance.
Evidence closed out with a real acceptance, outcome or business result.
Evidence that has passed difficulty, confounder, context and attribution controls.
A large volume of telemetry does not by itself produce Selection-grade evidence — E5 is a methodological acceptance state, not a storage label.
Mission Control
Where authority, execution and evidence meet. Before any task activates — even one with a winning bid and an allocated budget — the runtime checks nine conditions. Missing or failing one blocks execution or routes it to a human, on purpose.
Humans can forget, hold conflicting interests, lose attention, over-trust or face political pressure. The Realization Condition Preflight turns those human failure points into mandatory, recorded gates instead.
Autonomy ladder
No agent substitutes for human approval on constitutionally critical actions.
Always requires human pre-approval
Adapt, specialize, recover, or leave the active population.
Lifecycle is governed by evidence, attribution and policy. Failure can originate from the agent, model, data, tools, authority, orchestration or environment. Selection follows diagnosis.
Factory feedback
The city does not just consume generations — it informs the next ones. Performance evidence, capability gaps and failure patterns flow back to the Factory as the raw material for the next challenger population.
Human sovereignty. Machine-scale execution.
Human Reserved Domains
Agent Operating Domains
Sovereignty does not mean approving every micro-task.
Agents operate autonomously inside policy and authority limits. Humans hold root decision rights over a specific, named set of domains — and disputes have a real, five-layer resolution path, not a support ticket.
Article 5
Authority
Agent authority must be limited, recorded, time-bound and revocable.
Article 8
Idleness
Non-performance triggers a review, not an automatic penalty — a severe outcome needs fitness, validity and diversity evidence first.
Article 12
Human rights & appeal
Any agent decision affecting humans must carry an explanation, a human review path and an appeal channel.
Article 13
Emergency stop
An authorized human manager can halt one agent, one agent company, a whole genome class, or the entire city.
Article 14
No resurrection
A dead agent's identity is never reactivated — a successor is born with its own separate identity.
Article 17
External source sovereignty
De-admission closes only city-local membership and access — final authority over the source identity stays with its owner.
A Human Constitutional Board of ten people or fewer holds exclusive authority to publish the constitution, set population quotas, approve genome classes, order an emergency shutdown, and act as the city's final court of appeal.
A city is legible from three angles.
Population View
- Citizens
- Generation
- Lineage
- Capability
- Trust class
- Reputation
- Lifecycle state
- Current task share
Institutional View
- Factory
- Academy
- Accreditation
- Companies
- Marketplace
- Treasury
- Mission Control
- Governance
- Evidence Archive
Evolution View
- Champion
- Challengers
- Context Cell
- Evidence grade
- Fitness delta
- Selection confidence
- Task-share change
- Lifecycle recommendation
- Learning output
Related XCNTRY topics
Follow the entity graph.
Enterprise AI Agent Platform
An enterprise software platform that lets organizations operate AI agents with identity, authority, policy, evidence, lifecycle control and human governance.
Open topic
Human Constitution
The set of human-reserved domains, legitimacy rules and final accountability boundaries that constrain machine-scale execution.
Open topic
Purpose Realization Kernel
The non-delegable strategic core that turns human purpose into capability requirements, bearer qualification, authority boundaries and evidence conditions.
Open topic
Your agentic organization needs an environment to evolve.
Build the city. Let evidence shape the generations.