AI 提示詞： 客服自動化

Build a comprehensive FAQ section for [product/service/website].

What we do: [describe your offering]
Target audience: [who uses your product]
Common support tickets: [top 10 questions you receive]
Existing FAQ: [paste current FAQ, or 'none']
Tone: [professional / friendly / casual]

Create:
1. 20 FAQs organized into 4-5 logical categories
2. Each answer: clear, concise (under 100 words), with a specific action step
3. 5 'hidden' questions customers think but don't ask (and answer those too)
4. Internal links: which FAQ answers should link to other pages
5. A 'still need help?' CTA strategy for each category
6. SEO-optimized question phrasing that matches how people actually search

Design a chatbot flow for [purpose: customer support / lead qualification / order status / FAQ].

Platform: [website / app / WhatsApp / Facebook Messenger]
Top 5 customer intents: [what people ask most]
Escalation needed for: [when should a human take over]
Brand personality: [how the bot should 'sound']
Integrations available: [CRM, order system, knowledge base]

Build the chatbot:
1. A greeting message that sets expectations (what the bot can and can't do)
2. Intent recognition: 5-7 main conversation branches
3. Full conversation flow for each branch (with decision trees)
4. Fallback responses: what to say when the bot doesn't understand
5. Human handoff triggers and transition messages
6. A personality guide: do's and don'ts for the bot's communication style

Help me build a customer-facing knowledge base for [product/service].

Current documentation: [what exists now, or nothing]
Top support categories: [main topic areas]
User technical level: [beginner / intermediate / advanced / mixed]
Format preferences: [text, screenshots, video, GIFs]
Platform: [Zendesk, Intercom, Notion, custom]

Design the knowledge base:
1. Information architecture: categories, subcategories, and article hierarchy
2. 10 must-have articles (titles and outlines for each)
3. Article template: standard format every article should follow
4. Search optimization: how to title and tag articles for findability
5. A feedback mechanism: how to know which articles are actually helpful
6. A maintenance schedule: how often to review and update content

Build a library of canned responses for our support team.

Product/service: [what you support]
Channels: [email, chat, social, phone]
Team size: [number of agents]
Common scenarios: [list 10-15 frequent situations]
Brand voice: [professional / casual / empathetic]

For each scenario, create:
1. A canned response with [customization brackets] for personalization
2. A shorter version for chat (under 50 words)
3. A longer version for email (100-150 words)
4. Agent instructions: when to use this and when NOT to
5. Personalization tips: what details to add before sending
6. A naming convention so agents can find responses quickly

Help me build a ticket triage system for our support team.

Ticket volume: [daily/weekly count]
Team structure: [agents, specialists, tiers]
Categories: [list current categories, or 'need to define']
Priority levels: [how you currently prioritize, or 'need a system']
SLA requirements: [response and resolution time targets]

Design a triage system:
1. Category taxonomy: 6-10 categories with clear definitions
2. Priority matrix: how to assign P1/P2/P3/P4 based on impact and urgency
3. Routing rules: which tickets go to which team/specialist
4. Auto-tagging keywords: patterns that indicate category and priority
5. First-response templates for each priority level
6. A dashboard view: what metrics to track for triage effectiveness

Help me build a multi-layer support resolution system.

Product/service: [what you support]
Current support structure: [tiers, team size, tools]
Ticket volume: [daily/weekly]
Average resolution time: [current metrics]
Top 10 issue types: [list with approximate frequency]
Cost per ticket: [if known, by tier]

Design a multi-layer resolution engine:
1. Layer 0 (Self-Service): which issues can be fully resolved without human contact? Design the automation for each
2. Layer 1 (Frontline): which issues need a human but can be resolved in one touch? Create resolution scripts
3. Layer 2 (Specialist): which issues require deep expertise? Define routing criteria and knowledge requirements
4. Layer 3 (Engineering/Escalation): which issues need code changes or executive decisions? Build the handoff process
5. Smart routing logic: how to detect which layer an issue belongs to before a human reads it
6. Layer efficiency metrics: target resolution rate, time, and cost for each layer with improvement benchmarks

Help me design a self-service support experience for [product/service].

Current self-service: [what exists now]
Top tasks customers need help with: [list 5-10 common tasks]
Customer tech savviness: [beginner / intermediate / advanced]
Support cost per ticket: [if known]
Goal: [reduce ticket volume by X% / improve satisfaction]

Design the self-service experience:
1. A self-service homepage layout with smart search and popular topics
2. Interactive troubleshooting wizards for the top 3 issues
3. A decision tree: self-service vs. contact support (when each is appropriate)
4. Video tutorial outlines for visual learners (5 most common tasks)
5. A community forum structure where customers help each other
6. Metrics to measure self-service success: deflection rate, satisfaction, and completion rate