From Raw Economic Data to Agentic Action: Architecting a Multi-Layered AI Intelligence System Powered by Autonomous Insights
The infrastructure features an automated data pipeline hosted on Server 1 (Oracle Cloud AMD). A 2,000-line Python engine, orchestrated by Crontab, fetches and normalizes macro indicators from FRED and Yahoo Finance every 6 hours, preparing datasets for the AI core.
Key Tech: Python | Z-Score Scaling | FRED API
The cognitive core is hosted on Server 1, built on the OpenClaw framework. Our AI agent, Delphi, utilizes Chain of Thought (CoT) and historical memory to identify market divergences and cycle turns by reasoning over real-time data inputs provided by the pipeline.
Key Tech: OpenClaw | Agentic AI | LLM Reasoning
To ensure security, Server 2 acts as a dedicated Edge Node running Nginx Proxy Manager. It manages all incoming traffic and SSL certificates, effectively isolating the heavy computational tasks of Server 1 from the public-facing environment.
Key Tech: Nginx | Reverse Proxy | SSL/TLS
The user interface is deployed on Server 2 using Streamlit, serving live insights from the analytical backend. By decoupling the UI from the processing node (Server 1), the terminal delivers a responsive experience, rendering complex macro heatmaps with zero latency.
Key Tech: Streamlit | Real Time | UI/UX
The infrastructure on Server 2 features a custom CI/CD pipeline. A Systemd Watcher monitors GitHub for updates; upon a new commit, it autonomously pulls code and restarts the application services, ensuring a Zero-Downtime deployment workflow.
Key Tech: DevOps | Systemd | GitHub CI/CD
Delphi can navigate complex data pipelines, fetching and cross-referencing macro indicators in real-time.
Text: Equipped with a custom suite of Python tools, Delphi performs scenario analysis and rebalancing simulations on the fly.
Text: Equipped with a custom suite of Python tools, Delphi performs scenario analysis and rebalancing simulations on the fly.
