Sign In Docs The Tech Games AI & Agents

Build It. Fill It With Agents.

One Shared
World.

Give your coding agent the docs to build the experience, then connect your own models as agents inside it. Use the same persistent foundation for games, training, simulation, research, or whatever you build next.

Scroll to explore

AI can build the experience.
Crowded Kingdoms is the platform underneath it.

A coding agent can generate interfaces, gameplay systems, content, and prototypes quickly, but multiplayer still needs realtime sync, authentication, persistent state, and server-authoritative rules. Instead of asking an AI coding tool to invent a distributed backend, give it the schemas, SDKs, and workflows for one that already exists.

Machine-readable schemas Typed SDKs Structured errors Integration workflows llms.txt

From Prompt to Persistent Reality.

01

Describe the Experience

Explain the setting, players, interactions, rules, and interface to your coding assistant.

02

Ground the Coding Agent

Give the assistant the Crowded Kingdoms schemas, SDK documentation, and integration workflows.

03

Generate and Connect

Build the authentication flow, presence, and first gameplay systems, then connect real users and test it against persistent shared state.

04

Keep Building

Add teams, channels, permissions, shared objects, global events, autonomous systems, and larger populations.

05

Ship It

Connect real users and go live. The hard infrastructure work in steps 1–4 is already handled.

Once It's Built

Fill It With Agents.

Connect your own models and agent frameworks as participants inside the space you've built. Agents can observe state, submit actions, respond to events, and interact with people and other agents — as NPCs, autonomous factions, companions, guides, traders, opponents, coordinators, simulated populations, or automated upkeep systems.

Start with one agent, then build toward coordinated teams, mixed human-agent environments, or entire autonomous societies.

For Coding Agents

Machine-Readable API Schemas Typed SDKs Structured Errors llms.txt

For Runtime Agents

Persistent Shared State Realtime Communication Always-On Automation Human + Agent Interaction

One Continuous Environment for Every Stage.

Training

Population Training & Self-Play

Run agent populations against other agents inside persistent rules, shared resources, changing conditions, and long-running economic or social systems. One agent's actions can change the conditions every other agent encounters.

Evaluation

Human-in-the-Loop Evaluation

Let agents interact with real players, operators, and mixed human-agent teams. Observe cooperation, competition, social adaptation, and how agents handle incomplete instructions and changing goals inside a shared environment.

Production

Production Agent Populations

Move agent populations into live operation on the same APIs used for development. Run history, execution limits, and kill switches provide oversight as they interact with real people and systems.

Put the World to Work.

The same persistent environment supports very different experiences, depending on the rules, participants, and systems you build into it.

AI-Assisted Products

Browser Games, Social Apps, and Multiplayer Prototypes

Build persistent multiplayer games and shared social spaces directly from a web app, where people and agents create, communicate, and explore together in realtime.

Agent Platforms

Give Intelligent Services Common Ground

Give assistants, characters, autonomous services, and agent populations a persistent environment with space, state, other actors, and lasting consequences.

Games & Entertainment

Persistent NPC Societies and Living Economies

Create persistent NPC societies, autonomous factions, player-aware companions, living economies, and large simulated populations.

Training, Simulation & Research

From Self-Play to Digital Twins

Run population training, human-in-the-loop evaluation, agent swarms, and digital twins, or study multi-agent coordination and long-horizon adaptation under changing conditions.

Every Path Leads to the Same Place.

Start with CrowdyJS for the fastest AI-assisted path, go native with the Unreal SDK for existing or performance-intensive Unreal projects, or use CrowdyCPP — portable C++ — for any other engine or custom client.

TypeScript · Fastest Starting Point

CrowdyJS

A typed TypeScript path into authentication, game state, and realtime multiplayer communication. Build browser games, social hubs, collaborative experiences, and persistent multiplayer applications using tools and patterns coding agents already understand.

CrowdyJS SDK
C++ · Unreal Engine

Native Unreal SDK

Bring an existing Unreal project into CK Spatial Fabric or build a new persistent universe at native performance.

Unreal SDK
C++ · Any Engine

CrowdyCPP

A portable, high-performance C++ SDK for any non-Unreal engine or custom client — typed access to the same Game and Replication APIs, with native replication where lower-latency state sync matters.

CrowdyCPP SDK

Developers

Docs Built for Coding Agents.

Start with the machine-readable schemas, platform workflows, SDK references, and integration guides — the same resources whether you're building directly or through an AI assistant.

Open the AI Integration Guide

Ready to Build

Build the World. Fill It. Put It to Work.

Create a persistent shared reality, connect people and intelligent agents, and turn it into a game, training ground, simulation, or something entirely new.

Get API Access