Wie ein Mitarbeiter von Erewhon einen KI-gestützten Workflow entwickelte, der den Kundenservice in 10 Geschäften automatisiert

Herausforderung

Erewhon operates 10 store locations across Los Angeles, each with its own customer service inbox in HelpScout. Each store handled every customer service request manually, with no way to prioritize high-value members or maintain consistent response quality. The team was spending 20+ hours per week on repetitive ticket responses.

Lösung

Chris Morrison, Erewhon's Business Analyst and self-taught AI lead, built a 39-step Zap connecting Help Scout, ChatGPT, a vector store of institutional knowledge, and BigQuery. When a ticket arrives, the system:

• Checks the customer's membership status against Erewhon's database

• Routes members to a dedicated AI with a warmer tone; non-members get standard responses

• Drafts a reply using institutional knowledge about membership policies

• Surfaces the customer's purchase history so store managers can prioritize high-spend members

• Puts the crafted response directly into the hands of customer service to respond quickly and effectively

To prove it worked, Chris built a second AI: an "English Tutor" that grades each AI draft against the final human response, scoring how much the agent changed.

Befund

• 70 percent of AI-drafted responses sent without human modification

• 1,500 labor hours saved per year on customer service

• ~$40K/year in equivalent labor cost savings