Luminary Collective

Luminary Collective came to us with a problem that looks simple from the outside but is operationally complex at scale: four warehouses, three disconnected systems, and a reporting process that required two analysts working full days every week just to produce numbers that were already 48 hours stale by the time leadership saw them. The waste wasn't just in the hours spent — it was in every decision made without current information.

We started with the data infrastructure. The four warehouse management systems were feeding into nothing coherent. We built a unified Supabase data warehouse via n8n pipelines, establishing field-level ownership across Shopify, NetSuite, and Klaviyo to eliminate the conflicting records that had required manual reconciliation for years. Within the first week of go-live, the executive team had live inventory, sales, and fulfilment views for the first time.

The forecasting layer followed. We trained a multi-variate model on 24 months of historical sales velocity, seasonal patterns, and supplier lead times. It runs nightly, updating reorder recommendations across all SKUs. In the first 60 days, it benchmarked at 3.2× the accuracy of the prior human-managed process — not because the team was poor at their jobs, but because no human can hold 60 days of cross-SKU data in their head simultaneously.

The customer communication pipeline was the final piece. We replaced four static email templates with an LLM layer that generates context-aware messages based on purchase history, browsing behaviour, and support history. Messages pass through a lightweight tone classifier before being sent via Klaviyo — fully automated, fully auditable, and producing a 4× increase in communication volume without a single additional hire.