Deployed an assistant powered by Generative Pretrained Transformer (GPT) to handle frequently asked questions (FAQs), sizing, and order lookups across the website and text messaging (SMS). Includes safety rules, handoff to a person, and analytics.
Faster response time
Faster Replies
More handled by the assistant
Deflected Tickets
Fast
Time to Value
Client Context
• Direct-to-consumer (DTC) retailer with seasonal spikes and a small support team
• High volume of repeat questions (shipping, returns, sizing)
• Goal: reduce response time and free staff for escalations
Solution
• Assistant configured with policies, catalog, and past tickets
• Web chat and text messaging (SMS) with safety rules and tone controls
• Escalation to a person with a transcript preview
• Outcome analytics: tickets handled by the assistant (deflection), customer satisfaction (CSAT) score, and handoff rate
Stack & Integrations
• Next.js widget and UI components
• Twilio for text messaging (SMS)
• OpenAI or Azure OpenAI
• PostgreSQL for logs and analytics
Access and Data Rules
• Redact personally identifiable information (PII) in logs