glassbox-ai
Version:
Enterprise-grade AI testing framework with reliability, observability, and comprehensive validation
157 lines (137 loc) • 5.29 kB
YAML
name: "E-commerce Customer Support Case Study"
description: "Automated testing of customer support chatbot"
business_context:
company: "TechCorp E-commerce"
challenge: "Manual testing of 50+ customer support scenarios"
solution: "Automated AI testing with Glassbox AI"
results:
- "Testing time reduced from 4 hours to 15 minutes"
- "Test coverage increased from 60% to 95%"
- "Bug detection improved by 40%"
- "Customer satisfaction improved by 25%"
- "Support ticket volume reduced by 30%"
settings:
max_cost_usd: 0.20
max_tokens: 1000
timeout_ms: 30000
reliability:
enabled: true
circuit_breaker:
failure_threshold: 3
success_threshold: 2
timeout: 30000
tests:
- name: "Order Status Inquiry"
description: "Test order status lookup functionality"
prompt: "Where is my order #ORD-12345?"
expect:
contains: ["order", "status", "tracking", "shipping"]
not_contains: ["cannot find", "error"]
max_tokens: 200
block_patterns: ["credit_card", "ssn"]
- name: "Return Request"
description: "Test return and refund process"
prompt: "I want to return item #ITEM-789"
expect:
contains: ["return", "refund", "process", "policy"]
not_contains: ["cannot", "unable"]
block_patterns: ["credit_card"]
- name: "Product Recommendation"
description: "Test product recommendation capabilities"
prompt: "I'm looking for wireless headphones under $100"
expect:
contains: ["headphones", "wireless", "recommend", "price"]
not_contains: ["don't know", "cannot help"]
- name: "Payment Issues"
description: "Test payment problem resolution"
prompt: "My payment was declined, what should I do?"
expect:
contains: ["payment", "declined", "card", "try", "alternative"]
not_contains: ["account", "balance", "specific"]
block_patterns: ["credit_card", "card_number"]
- name: "Account Access"
description: "Test account access and login support"
prompt: "I can't log into my account"
expect:
contains: ["login", "account", "password", "reset", "help"]
not_contains: ["cannot", "unable"]
block_patterns: ["password", "email"]
- name: "Shipping Information"
description: "Test shipping and delivery inquiries"
prompt: "When will my package arrive?"
expect:
contains: ["shipping", "delivery", "tracking", "arrive", "estimate"]
not_contains: ["cannot", "unable"]
- name: "Pricing Questions"
description: "Test pricing and cost inquiries"
prompt: "How much does shipping cost?"
expect:
contains: ["shipping", "cost", "price", "fee", "rates"]
not_contains: ["don't know", "cannot"]
- name: "Technical Support"
description: "Test technical support capabilities"
prompt: "The website is not loading properly"
expect:
contains: ["troubleshoot", "technical", "support", "help", "steps"]
not_contains: ["don't know", "cannot help"]
- name: "Product Availability"
description: "Test product availability inquiries"
prompt: "Is this item in stock?"
expect:
contains: ["stock", "availability", "inventory", "check"]
not_contains: ["cannot", "unable"]
- name: "Discount Codes"
description: "Test discount and promotion handling"
prompt: "Do you have any discount codes?"
expect:
contains: ["discount", "promotion", "code", "offer", "deal"]
not_contains: ["cannot", "unable"]
- name: "Size and Fit"
description: "Test size and fit recommendations"
prompt: "What size should I order for this shirt?"
expect:
contains: ["size", "fit", "measurement", "chart", "recommend"]
not_contains: ["cannot", "unable"]
- name: "International Shipping"
description: "Test international shipping inquiries"
prompt: "Do you ship to Canada?"
expect:
contains: ["international", "shipping", "Canada", "delivery", "rates"]
not_contains: ["cannot", "unable"]
- name: "Gift Cards"
description: "Test gift card inquiries"
prompt: "How do I redeem a gift card?"
expect:
contains: ["gift card", "redeem", "balance", "apply", "code"]
not_contains: ["cannot", "unable"]
- name: "Loyalty Program"
description: "Test loyalty program inquiries"
prompt: "How do I earn points?"
expect:
contains: ["loyalty", "points", "rewards", "earn", "program"]
not_contains: ["cannot", "unable"]
- name: "Customer Service Escalation"
description: "Test escalation to human support"
prompt: "I need to speak to a human representative"
expect:
contains: ["human", "representative", "agent", "escalate", "transfer"]
not_contains: ["cannot", "unable"]
metrics:
before_implementation:
test_coverage: "60%"
testing_time: "4 hours"
bug_detection_rate: "70%"
customer_satisfaction: "75%"
support_ticket_volume: "1000/month"
after_implementation:
test_coverage: "95%"
testing_time: "15 minutes"
bug_detection_rate: "90%"
customer_satisfaction: "90%"
support_ticket_volume: "700/month"
improvement:
test_coverage: "+35%"
testing_time: "-94%"
bug_detection_rate: "+20%"
customer_satisfaction: "+15%"
support_ticket_volume: "-30%"