The Inspiration: A Business Problem That Needed a Better Solution Running an online arbitrage operation means making dozens of purchasing decisions every day. Each one depends on sales data, margin analysis, market trends, and risk factors. I was doing all of that manually, and I realized the logic behind each decision was actually consistent and repeatable. If the criteria were consistent, a system could make the call. I set out to build it.

The Objective: Encoding Business Logic into a System My goal was not to build something flashy. It was to take the exact decision making process I was already using and encode it into a pipeline that could run without me. Every buy or pass decision had rules behind it. I wanted those rules to live in a system, not in my head.

The Architecture: Deterministic Rules Meet AI Judgment The system operates in two layers. The first layer is fully deterministic. It pulls marketplace and market data through APIs and runs every product through a strict set of weighted criteria around margin, sales velocity, competition, and risk. Most decisions get made here. The second layer handles the edge cases that rules alone cannot resolve. An LLM analyzes ambiguous signals like trend anomalies or intellectual property risk and returns a structured judgment that feeds back into the pipeline. Everything is auditable.

Current Status

• Phase: Active Development

• Status: In Progress. Core deterministic pipeline is built. LLM integration layer is currently being refined and tested.