// Technical architecture

AULONG uses Brain, Data, Skill, Risk Guard, Chain and Open API to form a multi-layered agent architecture, connecting AI capabilities, financial data, risk control and open interfaces in the same system.

// Capability stack
Six-layer stack
AULONG's technical route is to integrate MexQuant Capital Ltd's capabilities in AI quantitative research, financial data processing, risk identification and system architecture into a multi-agent collaboration network.
01
Data layer
Access market conditions, on-chain data, news events, macro variables and user tasks to complete collection, cleaning and structured processing.
02
Brain Intelligent hub layer
Understand user goals, dismantle task paths, and schedule different agents to collaboratively complete complex tasks.
03
Agent collaboration layer
Different Agents collaborate around research, analysis, risk control, execution and interaction.
04
Skill module layer
Carrying capabilities such as market analysis, strategy research, report generation, API calls and life services.
05
Risk control layer
Monitor market anomalies, data fluctuations, mission authority and execution risks.
06
Open API ecological layer
Open interfaces for developers, institutions and ecological applications, supporting task calls, Skill services and AUL consumption settlement.
AULONG's agent safety boundary
Data isolation
Maintain data permission isolation between different users, tasks and modules.
Least privilege
The agent is given only the necessary permissions required to complete the current task.
Tool whitelist
High-risk tools, interfaces and on-chain operations must be reviewed before being called.
Risk Guard Monitoring
Continuously identify unusual signals, task deviations, and high-risk behaviors.
Traces on the chain
Key tasks, call records and settlement behaviors can be recorded on-chain.
Abnormal fuse
Automatically suspend tasks when abnormal instructions, data distortion, or risk expansion are discovered.
// Technical route
Technical route in one line
AULONG does not allow AI to only answer questions, but uses Data, Brain, Agent, Skill, Risk Guard and Open API to Empower AI Agents to understand tasks, invoke tools, identify risks, and execute collaboratively.
View technical roadmap