When this matters. This matters when you want categories and AI-assisted routing to reflect how your team actually works. Start with the categories, then write the business context around those categories.
How to set it up. First define the conversation types you want separated. For a plumbing company, useful categories might be urgent_issue, quote_request, scheduling, billing_payment, and service_details. Then write the business context as a short job description for the workspace: “Business texting app for a plumbing company handling urgent plumbing issues, quote requests, appointment scheduling, billing or payment questions, and general service detail inquiries.”
Good example. This structure works because it gives DribX a clean map. “Business texting app” sets the channel. “Plumbing company” sets the industry. The category list tells it which intent buckets matter. The category guidance then explains the boundaries: “Classify the dominant plumbing intent. Questions about hours, weekend service, service area, capabilities, or what is included should be service_details. Concrete booked time slots should be scheduling.”
What happens afterward. When a customer sends an incoming message, DribX can compare the text against both the business context and the category boundaries. “Do you come on weekends?” can stay in service_details. “My appointment is Saturday at 10, can we move it?” can become scheduling. “Water is coming into my basement” can become urgent_issue.
Common mistake to avoid. Avoid treating each category as a separate mini-description with no relationship to the others. The hard part is often not defining one category; it is explaining how to choose between two categories when a message could almost fit both.