Xcity Agentic Platform

AI Agent as a service.

Three-layer AI platform built on solar energy. From multi-engine orchestration to model routing to agent execution — powered by 100GW of clean energy.

Xcity Builder Xcity Exchange Xcity Agent

Three Layers, One Platform

The Xcity Agentic Platform spans three architectural layers. Each layer is independently powerful, and together they form a complete stack from orchestration to execution.

01
L1

Xcity Builder

Multi-engine orchestration + dynamic task routing + MCP context protocol

4 Engines
02
L2

Xcity Exchange

Model Hub + Data Hub + Compute Hub + Energy Hub — four-factor trading bus

4 Hubs
03
L3

Xcity Agent

Long-term memory + cross-model sessions + persistent context + human review

∞ Sessions

Four callable capabilities

Discover Models, Skills, MCPs and Agents via GET /v1/capabilities. Call everything from the same SDK.

Models

250+ models, one API

Skills

Server-side prompt engine

MCPs

Gateway-proxied tools

Agents

A2A runtime
L1 — Builder Layer

Xcity Builder:
Multi-Engine Orchestration

Don't choose one framework. Let the platform route tasks to the best engine for each job. Xcity Builder automatically selects between LangGraph, CrewAI, n8n, and Dify based on task complexity and structure.

  • Dynamic task routing — no manual engine selection
  • MCP (Model Context Protocol) for standardized context
  • Supports LangGraph, CrewAI, n8n, Dify out of the box
  • Extensible engine adapter interface
LangGraphCrewAIn8nDify
Task Routing Flow
INPUT TASK
{ type: "branching", steps: 12 }
Xcity Router
analyzes → routes → executes
LangGraph
complex
CrewAI
multi-agent
n8n
workflows
Dify
simple
MCP standardized context → output
Exchange
Bus
Model Hub
250+ models
Data Hub
Ocean C2D
Compute Hub
GPU Clusters
Energy Hub
100GW Solar
L2 — Exchange Layer

Xcity Exchange:
Four-Factor Trading Bus

A unified exchange for Model, Data, Compute, and Energy. Built on distributed bus architecture, Xcity Exchange connects AI workloads to the right resources at the right price — all powered by 100GW of solar capacity.

  • Model Hub — LiteLLM-powered multi-model routing (GPT, Claude, Gemini, Llama, Mistral)
  • Data Hub — Ocean Protocol C2D for privacy-preserving data exchange
  • Compute Hub — GPU cluster allocation for AI training and inference
  • Energy Hub — Solar compute tokens backed by 100GW capacity
🤖
250+

Model Hub

250+ models via LiteLLM

🗄️
C2D

Data Hub

Ocean C2D privacy compute

⚡
GPU

Compute Hub

GPU cluster orchestration

☀️
100GW

Energy Hub

100GW solar capacity

L3 — Agent Layer

Xcity Agent:
Long-Term Memory & Sessions

Agents that remember. Xcity Agent combines persistent long-term memory with cross-model session management — enabling AI agents that build context over time, across sessions, and across models.

  • Long-term memory with vector search and structured recall
  • Cross-model session continuity (GPT ↔ Claude ↔ Gemini)
  • Persistent context across sessions and devices
  • Human-in-the-loop review for high-stakes decisions
Read the Docs →
Agent Memory Graph
MEM HUM GPT CLD VDB SES Xcity
Xcity
GPT
Claude
VDB
SES
HUM
Get Started

Ready to Build on
Xcity Agentic Platform?

Access AI models, orchestration tools, and solar-powered compute — all in one platform built for the AI city-state.

Read the Docs Solar Compute