FROM N8N TO OPENCLAW

Wander Tiger started as an attempt to reduce weekend research work with automation. It gradually evolved into a set of agentic AI experiments across trading, research, banking, data sources, and cloud infrastructure.

TIMELINE

n8n on Pikapods: The first setup used n8n hosted on Pikapods for option and stock screening, research workflows, and weekend research preparation. Automation helped, but human review and manual weekend work were still required.

Manual Bottleneck: With two young kids, the weekend operating burden became difficult to sustain. The problem was no longer just automation. The workflow needed more independent reasoning, monitoring, and execution support.

OpenClaw Experiment: When OpenClaw started gaining momentum in China, I began experimenting with it locally on my M1 Pro MacBook. This marked the shift from low-code automation toward agentic AI workflows.

Athena on DigitalOcean: Athena became the first OpenClaw agent. She was initially hosted on a DigitalOcean Droplet and focused on option trading, screening, and risk-defined strategy workflows.

More Agents on Hetzner: As my confidence and knowledge of OpenClaw grew, I started building more agents. Minerva, the Personal Banker agent, is hosted on Hetzner. I considered Alibaba Cloud, but the account opening process was too troublesome to complete.

DATA SOURCES

Started with yfinance for market data experiments.

Moved to Marketdata.app for options and market data workflows.

Also uses Alpha Vantage API as part of the data stack.

Stock screening started on FinViz and recently moved to TradingView.

LESSONS

Automation is useful, but manual review can still become the real bottleneck.

Agentic systems become more valuable when they reduce weekend maintenance, not just clicking.

Hosting choices matter: local machines are good for experiments, cloud servers are better for always-on agents.

Data sources evolve as the project moves from prototypes toward operating workflows.