A UK-based Amazon Merch seller with 6,000+ product designs was generating half a million in annual revenue — but a significant portion was being consumed by ad costs with no clear visibility into which products justified the spend.

The core metric wasn't ACoS (the industry standard). It was Royalty Per Ad Dollar — how much actual royalty income each advertising dollar generated. At a target of 24%, most products were falling short, but there was no system to identify which ones or recommend corrective action.

The 3-person team was spending hours manually checking spreadsheets. They needed a system that would watch everything, learn patterns, and surface decisions — not just data.

Rather than building a monolithic dashboard, I designed a multi-agent architecture where each agent owns one responsibility — but they all share context through a structured observation log in Supabase.

This means the Ad Waste Detector doesn't just flag underperformers — it writes a structured observation that the Pricing Optimizer can read and act on. The system develops institutional knowledge rather than just executing tasks.

Within the first week of deployment, the system identified over $30,000 in wasted ad spend across underperforming products — spend that had been invisible in manual spreadsheet reviews.

The CEO Brief agent now delivers a weekly executive summary directly to Slack, written in natural language with specific recommendations. The team went from spending hours on data analysis to making decisions in minutes.

Most importantly, the system learns. Each agent's observations become context for every other agent. The Pricing Optimizer reads what the Ad Waste Detector found. The Trend Analyzer informs the CEO Brief. It's not nine separate tools — it's one system that thinks.