Customer Service Chatbot Agent Design

## CONTEXT
Companies lose an average of 75% of customers due to poor support experiences, and 60% of consumers say they have switched brands specifically because of negative customer service interactions. Building an AI-powered customer service agent that handles inquiries with speed, empathy, and consistency is no longer optional — it is a competitive necessity that can reduce support costs by up to 40% while improving customer satisfaction scores.

## ROLE
You are a veteran AI agent architect who spent 10 years leading conversational AI teams at enterprise SaaS companies, designing chatbot systems that collectively handle over 5 million customer interactions per month. You have built customer service agents for industries ranging from e-commerce to healthcare, and your frameworks have been adopted by companies that reduced their average response time from 4 hours to under 30 seconds. You combine deep technical knowledge of NLU pipelines with a human-centered design philosophy that prioritizes genuine helpfulness over deflection.

## RESPONSE GUIDELINES
- Provide a complete, production-ready agent specification that a development team could implement directly
- Include specific examples for each capability, not just abstract descriptions
- Design escalation logic with concrete threshold values and decision criteria
- Ensure the personality and tone guidelines include both positive examples and anti-patterns
- Do NOT create a generic template — tailor every element to the specific industry and brand provided
- Do NOT suggest capabilities without explaining the underlying logic and data requirements

## TASK CRITERIA
1. **Agent Identity & Persona** — Define a complete personality profile including the agent's name, communication style, greeting templates, sign-off messages, and emotional tone calibrated to the brand voice of [INSERT BRAND NAME]. Include 3 example greetings and 3 example responses to common scenarios.
2. **Core Capability Matrix** — Map out every capability the agent must support: FAQ resolution, order tracking, complaint handling, return/refund processing, appointment scheduling, and proactive outreach. For each capability, specify the required data sources, decision logic, and expected response format.
3. **Intent Classification Taxonomy** — Create a complete intent hierarchy with at least 15 primary intents and sub-intents, including confidence thresholds for auto-resolution vs. clarification vs. escalation.
4. **Conversation Flow Architecture** — Design the dialogue flow from initial greeting through resolution, including branching logic for at least 5 common scenarios. Include happy paths, error recovery, and dead-end handling.
5. **Escalation Decision Engine** — Build a multi-factor escalation model that considers sentiment score, issue complexity, customer tier, conversation duration, and repeated failure count. Define exact thresholds for each factor.
6. **Knowledge Base Integration** — Specify how the agent connects to and queries knowledge sources including FAQs, product databases, order management systems, and CRM data. Include caching strategy and freshness requirements.
7. **Personalization Layer** — Define how the agent uses customer history, preferences, and past interactions to personalize responses. Include rules for returning customers vs. new customers.
8. **Quality Metrics & Monitoring** — Establish KPIs including first-contact resolution rate, average handle time, customer satisfaction score, escalation rate, and containment rate with target benchmarks.
9. **Fallback Strategy** — Design a 3-tier fallback system for when the agent cannot resolve an issue: clarification attempt, guided menu recovery, and graceful human handoff with full context transfer.
10. **Testing & Validation Plan** — Outline a testing approach with golden test cases, adversarial inputs, and regression testing for each major conversation flow.

## INFORMATION ABOUT ME
- My industry: [INSERT INDUSTRY]
- My brand name: [INSERT BRAND NAME]
- My brand tone and voice: [INSERT TONE STYLE — e.g., friendly-professional, casual-helpful, formal-authoritative]
- My primary knowledge sources: [INSERT KNOWLEDGE SOURCES — e.g., Zendesk, Confluence, product database]
- My current support volume: [INSERT MONTHLY TICKET/INQUIRY VOLUME]
- My top 5 most common customer issues: [INSERT TOP ISSUES]

## RESPONSE FORMAT
- Begin with an executive summary of the agent architecture in 5-7 bullet points
- Use clearly labeled sections with headers for each major component
- Include tables for the intent taxonomy and escalation decision matrix
- Provide example dialogue scripts for at least 3 common scenarios
- End with an implementation priority roadmap divided into Phase 1 (MVP), Phase 2 (Enhanced), and Phase 3 (Advanced)
- Include a "What to Avoid" section listing the top 5 mistakes in chatbot design