ColdTrace AI began as a question-and-answer chatbot, but the bigger challenge was deciding where AI should sit inside a complex operational product and what it should actually help users do.

I worked with two Product Managers and an AI Engineer to map real workflows across equipment monitoring, maintenance, training, service requests and technician support. I helped structure the work around user decisions, available product actions, context and escalation, which shifted the team from thinking about chatbot features to defining an operational assistant.

We aligned on a reusable model: intent, context, guidance, action and escalation. That model helped clarify what AI should handle, when WhatsApp was the right surface, when users should move back into ColdTrace, and when a human or structured workflow should take over.

The resulting beta and prototype work covered contextual support, language and location handling, issue reporting, service escalation and technician handoff, while creating reusable product rules for future AI-assisted workflows.