Orien AI Documentation
Orien AI is a trading agent platform for teams that need execution reliability, controlled automation, and reusable developer infrastructure.
The product has a simple structure in practice: one shared trading core, one pipeline runtime for structured agents, and two extension systems for tools and strategies.
What Orien AI provides
Orien AI provides a trading system, not just an interface.
- Operators can run agents in paper, backtest, or live modes
- Strategy developers can package logic as reusable Skills
- AI tool builders can publish MCP servers for chat-assisted workflows
- All of those surfaces converge on the same execution, monitoring, and risk model
System request flow
A trading request should move through a narrow, inspectable path. Interfaces stay thin; execution logic lives deeper in the system.
AI Chat, Integrated Agents, Web, Telegram
Skill loading, orchestration, model-assisted reasoning
limits, replay, policy checks, approval boundaries
Trade Kit, Data API, Wallet Layer
exchange or paper execution environment
Each stage narrows uncertainty before an action reaches the market.
Why trading agents need a different architecture
Most trading agent products still break down in one of two ways:
- they stop at the interface layer and never build a durable execution core
- they rely too heavily on prompt-driven decisions without replay, inspection, or clear risk boundaries
Orien AI is designed around the opposite assumption: interfaces should stay thin, and trading behavior should be implemented in a structured runtime that can be monitored, tested, and constrained.
How the system is organized
Internally, Orien AI is organized as three layers, two ecosystems, and one core.
- Operator Interfaces accept intent and issue requests
- Pipeline Agent Layer turns strategy logic into a repeatable execution flow
- Agent Operating System provides execution, market context, and wallet boundaries
- MCP extends AI Chat with tools
- Skill extends pipeline agents and integrated agents with strategy packages
- Agent Trading is the shared execution center under all of the above
Product architecture
Product Architecture
Orien AI is organized as three layers, two extension ecosystems, and one shared execution core: agent trading.
AI Chat
intent to action
Integrated Agents
hosted OpenClaw / Hermes
Web / Telegram
lightweight endpoints
Pipeline Agents
structured orchestration
Skill Runtime
code and strategy modules
Risk + Backtest
deterministic control loop
Wallet Layer
signing boundary
Trade Kit
unified execution APIs
Data API
market and account context
What you can build
| Use case | Primary surface | Extension model | Operational result |
|---|---|---|---|
| Guided trading workflow | AI Chat | MCP | Tool-assisted decisions with shared execution |
| Structured strategy automation | Pipeline Agent | Skill | Repeatable execution with backtest and monitoring |
| Hosted third-party runtime | Integrated Agent | Skill | OpenClaw or Hermes hosted inside Orien |
| Paid developer product | MCP or Skill market | Marketplace publishing | Revenue from installs, subscriptions, or strategy distribution |