From a 3-week pilot to the knowledge layer of a global trade network.

A global trade and logistics operation runs across filing systems, TMS, internal spreadsheets, and the heads of senior analysts. Producing a single quarterly business review meant hours of manual reconciliation across mismatched sources, and customers waited days for answers the team couldn't assemble in real time. The data warehouse was there. The knowledge for interpreting it lived in the heads of the people who had spent years learning what the numbers meant. Off-the-shelf copilots produced plausible-sounding answers the senior analysts couldn't defend — fluent, confident, and wrong in the ways that mattered most.

Pedestal’s analyst connected to the team’s warehouse, mapped what was already there, and pinged the senior domain experts in the chat tools they already lived in. Their answers came back into Pedestal and were captured as versioned rules in Pedestal Atlas — the governed knowledge layer every Pedestal agent inherits from, with every rule tied back to the source that verified it. Over a few weeks, the analyst compressed years of accumulated expertise across the team’s core analytical workflows — the recurring questions, the investigative ones, and the judgment calls in between. Evaluated blind against the team’s legacy system of record, it matched expert answers at 95%+ accuracy.

What started inside the operator now runs across dozens of their downstream customers — the partners who rely on the platform for day-to-day operational decisions, visibility, and compliance. Quarterly review generation collapsed from a full day to minutes, and new agents — operational monitoring, exception handling, customer-facing coordination — inherit Atlas on day one. What had been years of senior-analyst experience, locked inside a handful of people, is now Atlas: the governed knowledge layer every downstream system runs on.