AI Agent Architecture

Why Python Agents?

Traditional Smart Contracts are passive; they cannot execute themselves or "think." While tools like Chainlink Automation trigger actions based on simple time/logic checks, they lack the ability to process complex data like Market Sentiment or News Analysis.

AUREO solves this by utilizing an off-chain Python Agent (main.py) that acts as the "brain," while the Smart Contract (AureoRWAPool) acts as the "muscle".

The Loop Architecture

1

The Monitoring Phase (Read)

The Agent runs an infinite loop (e.g., every 60 seconds) to fetch data from two sources:

• On-Chain Data: Calls getGoldPrice18Decimals() from the contract to get the true execution price.

• Off-Chain Data: (Planned) Calls Google Gemini API to analyze market sentiment text or chart patterns.

2

The Decision Phase (Think)

Instead of a simple "Limit Order," the Agent uses a composite score logic.

pseudo.py

# Pseudo-code logic for Agent Decision
def analyze_market(price, sentiment_score):
    is_dip = price < TARGET_PRICE
    is_safe = sentiment_score > 0.5 # Positive news environment

    if is_dip and is_safe:
        return "BUY"
    return "WAIT"