Insurance Agent Intelligent Trainer / 保险代理人智能陪练系统
English: AI-powered insurance agent coaching system — parses product documents, generates personalized question banks, assesses agent competency levels, schedules daily training based on client visits, and runs interactive role-play drills. Benchmarked against AIA, Ping An, and Alibaba Cloud insurance training systems.
中文: 保险代理人智能陪练系统——解析产品文档、自动生成问题库、评估代理人能力等级、 结合当日客户拜访行程安排个性化训练、进行情景对练。对标友邦保险、平安保险、阿里云智能陪练水平。
Trigger Keywords / 触发关键词
Immediately activate when user mentions:
- 保险陪练 / 产品陪练 / 智能陪练 / 代理人训练
- 代理人培训 / 新人培训 / 保险话术训练
- 产品演练 / 客户异议处理 / 保险销售训练
- agent training / insurance coaching / product drill / sales training
- skill assessment / competency evaluation / agent profiling
- daily training plan / training schedule / personalized coaching
Core System Architecture / 核心系统架构
┌─────────────────────────────────────────────────────────────────┐
│ Insurance Agent Intelligent Trainer │
├─────────────────────────────────────────────────────────────────┤
│ ┌──────────────┐ ┌──────────────┐ ┌──────────────────────┐ │
│ │ Product Doc │ │ Agent Profile│ │ Daily Schedule/Routes│ │
│ │ Parser │ │ Engine │ │ Integration │ │
│ │ (PDF/Word/ │ │ (Skill Level │ │ (Today's Visits & │ │
│ │ Images) │ │ Assessment) │ │ Client Profiles) │ │
│ └──────┬───────┘ └──────┬───────┘ └──────────┬───────────┘ │
│ │ │ │ │
│ ▼ ▼ ▼ │
│ ┌──────────────────────────────────────────────────────────┐ │
│ │ Question Bank Generation Engine │ │
│ │ Product Knowledge │ Objection Handling │ Case Analysis │ │
│ │ [5 difficulty tiers × 3 categories = 15 question types] │ │
│ └──────────────────────────┬───────────────────────────────┘ │
│ │ │
│ ▼ │
│ ┌──────────────────────────────────────────────────────────┐ │
│ │ Personalized Training Scheduler │ │
│ │ [Skill Level + Schedule + Product Priority = Daily Plan]│ │
│ └──────────────────────────┬───────────────────────────────┘ │
│ │ │
│ ▼ │
│ ┌──────────────────────────────────────────────────────────┐ │
│ │ Interactive Training Engine │ │
│ │ Role-play │ Real-time Feedback │ Progress Tracking │ │
│ └──────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────────┘
Core Capabilities / 核心能力
1. Product Document Parser / 产品文档解析引擎
Supported formats: PDF, Word (.docx), scanned images (with OCR), plain text
Parsing pipeline:
Document Upload
│
▼
[Format Detection] → PDF / Word / Image / Text
│
▼
[Text Extraction] → Raw text content
│
▼
[Structure Analysis]
├─ Product name, type, target customers
├─ Coverage scope (death, medical, annuity, critical illness, etc.)
├─ Premium levels & payment periods
├─ Policy terms & exclusions
├─ Sales pitch key points
├─ Competitive advantages vs. similar products
└─ Compliance notes & regulatory requirements
│
▼
[Structured Product Profile] → Ready for question generation
Output: Structured Product Profile JSON
{
"product_name": "XX福享人生终身寿险(万能型)",
"product_type": "whole-life insurance with universal account",
"insurer": "国联人寿",
"target_customers": ["30-50岁中高收入人群", "有财富传承需求"],
"coverage": {
"death_benefit": "100%-160%账户价值",
"annuity_option": "60岁起可转换为年金",
"waiver": "可选投保人保费豁免"
},
"premium": {
"min_annual": 12000,
"payment_periods": ["3年", "5年", "10年", "20年"],
"min_coverage_years": "终身"
},
"key_selling_points": [
"复利增值,万能账户历史结算利率4.5%-5.2%",
"灵活追加,额外资金可随时进入万能账户",
"身故保障与财富传承双重功能"
],
"competitive_edges": ["结算利率优于同类竞品", "追加无上限"],
"exclusions": ["投保人对被保险人的故意伤害", "2年内自杀(无民事行为能力人除外)"],
"compliance_notes": ["需双录(录音录像)", "犹豫期15天", "等待期90天"],
"difficulty_tags": ["新人友好", "需强化健康告知", "财务规划综合能力"]
}
2. Agent Profile & Skill Assessment / 代理人画像与能力评估
Three skill tiers:
| Tier | Level | Description | Training Focus |
|---|---|---|---|
| 🌱 L1 - 入门级 | Beginner | < 1 year experience, struggles with product details and objection handling | Foundation: product knowledge, basic sales scripts, simple objection responses |
| ⚡ L2 - 进阶级 | Intermediate | 1-3 years, solid product knowledge but inconsistent closing rate | Application: complex scenarios, multi-product combination, competitive replacement, high-net-worth clients |
| 🎯 L3 - 专家级 | Advanced | 3+ years, high performance, needs strategy for complex cases | Mastery: enterprise/group clients, tax planning, estate planning, competitive stealing, mentoring skills |
Profile structure:
{
"agent_id": "AG20240001",
"name": "张明",
"level": "L2",
"level_label": "进阶级",
"tenure_years": 2.5,
"certifications": ["保险代理人资格证", "健康险销售资质"],
"performance": {
"monthly_premium_target": 50000,
"monthly_premium_actual": 42000,
"closing_rate": 0.32,
"avg_policy_size": 18500,
"new_customer_rate": 0.45
},
"product_mastery": {
"term_life": 0.85,
"whole_life": 0.72,
"critical_illness": 0.58,
"medical_insurance": 0.80,
"annuity": 0.45,
"investment_linked": 0.38
},
"weak_points": [
"健康险异议处理不够熟练",
"不了解高端客户的税务筹划需求",
"组合产品销售话术单一"
],
"strong_points": [
"老客户维护能力强",
"缘故市场开拓优秀"
],
"daily_schedule": [
{"time": "09:00-10:00", "activity": "晨会", "location": "营业部"},
{"time": "10:30-12:00", "activity": "拜访客户A(国企中层,有养老需求)", "location": "客户公司"},
{"time": "14:00-15:30", "activity": "拜访客户B(私企业主,健康险需求)", "location": "客户公司"},
{"time": "16:00-17:30", "activity": "缘故客户C(教育金规划)", "location": "咖啡厅"}
]
}
3. Question Bank Generation / 问题库自动生成
Generated from product profile + agent level + training objectives
Question Types (15 categories across 3 dimensions)
By Category:
| Category | Description | Example |
|---|---|---|
| 产品知识 | Product features, terms, coverage | "XX福的等待期是多久?" |
| 客户画像 | Target customer identification | "什么样的客户适合购买这款产品?" |
| 异议处理 | Objection handling scripts | "客户说'我已经有社保了,不需要商业保险',如何回应?" |
| 案例分析 | Real case discussion | "40岁国企中层,年薪50万,如何用这款产品做养老规划?" |
| 合规话术 | Compliance-approved scripts | "如何向客户解释犹豫期和退保损失?" |
| 竞品对比 | vs. competitors | "相比平安福,这款产品的核心优势是什么?" |
| 促成话术 | Closing techniques | "客户表现出购买意向,如何自然促成?" |
| 交叉销售 | Multi-product combination | "如何将主险与医疗险组合销售?" |
By Difficulty (5 tiers):
| Level | Target Audience | Question Complexity |
|---|---|---|
| ⭐ 基础 | L1新人 | 单一产品,单一问题,直接答案 |
| ⭐⭐ 入门 | L1-L2 | 单一产品,1-2个知识点,需要解释 |
| ⭐⭐⭐ 进阶 | L2 | 单一产品,3-5个知识点,需组合分析 |
| ⭐⭐⭐⭐ 高阶 | L2-L3 | 多产品组合,竞争替换,高净值客户 |
| ⭐⭐⭐⭐⭐ 专家 | L3 | 综合方案,税务筹划,财富传承 |
Question Bank Generation Prompt:
Based on the product profile provided, generate a question bank with:
1. For each difficulty tier (基础/入门/进阶/高阶/专家):
- 5 multiple choice questions (产品知识)
- 3 case analysis questions
- 3 objection handling scenarios
- 2 competitive comparison questions
- 1 closing technique exercise
2. Total: 65+ questions per product
3. For each question, provide:
- Question text
- Difficulty level (1-5)
- Category (产品知识/异议处理/案例分析/竞品对比/促成话术)
- Ideal answer / model response
- Evaluation criteria (excellent/good/needs-improvement)
- Coaching tips for the trainer
4. Personalized Training Scheduler / 个性化训练调度引擎
Input factors:
Agent Profile (Level + Weak Points)
+
Today's Client Schedule (Who → What need → What product)
+
Product Priority Matrix
=
Personalized Daily Training Plan
Scheduling Algorithm:
def generate_daily_training_plan(agent_profile, daily_schedule, products):
"""
Generate personalized training plan for the day.
"""
# Step 1: Identify today's client visit products
today_products = extract_products_from_schedule(daily_schedule)
# Step 2: Get agent's weakness areas for these products
weakness_map = get_weakness_for_products(
agent_profile.weak_points,
today_products
)
# Step 3: Calculate training time available
available_minutes = calculate_available_training_time(daily_schedule)
# Step 4: Prioritize by impact × weakness × product value
training_queue = prioritize_training(
weakness_map,
today_products,
agent_profile.level,
time_constraint=available_minutes
)
# Step 5: Generate session plan
sessions = split_into_sessions(training_queue, available_minutes)
return {
"date": today,
"agent": agent_profile.name,
"total_minutes": available_minutes,
"sessions": sessions,
"focus_products": today_products,
"key_objectives": get_key_objectives(training_queue)
}
Example Daily Training Plan:
{
"date": "2026-05-05",
"agent": "张明",
"level": "L2",
"total_minutes": 90,
"sessions": [
{
"time": "08:00-08:20",
"duration": 20,
"type": "晨间快练",
"mode": "快问快答",
"focus": "年金险产品知识(高频问题5题)",
"product": "福享人生终身寿险",
"objective": "巩固年金转换权的计算逻辑"
},
{
"time": "12:30-13:00",
"duration": 30,
"type": "午间强化",
"mode": "情景对练",
"focus": "健康险异议处理",
"scenario": "客户:"我有社保,不需要商业医疗险"",
"product": "康健医疗保险",
"level": "⭐⭐⭐ 进阶",
"coaching_tips": "引导客户认识到社保报销比例上限,用自费药比例对比引发需求"
},
{
"time": "17:30-18:30",
"duration": 40,
"type": "晚间复盘",
"mode": "案例分析 + 角色扮演",
"focus": "私企业主综合保障方案",
"scenario": "45岁私企老板,年收入200万,已有多份保单,如何做加保方案?",
"products": ["终身寿险+万能账户", "高端医疗", "企业财产险"],
"level": "⭐⭐⭐⭐ 高阶",
"model_response_guide": "从家庭资产与企业资产隔离角度切入,引出终身寿险的债务隔离和传承功能"
}
],
"key_metrics_to_track": [
"异议处理响应时间(目标<30秒)",
"产品知识点正确率(目标>85%)",
"方案组合完整性(3单以上产品覆盖)"
]
}
5. Interactive Training Session / 智能陪练对话引擎
Session modes:
| Mode | Description | Duration | Best For |
|---|---|---|---|
| 快问快答 | Rapid-fire Q&A | 5-10 min | Pre-meeting warmup |
| 情景对练 | Role-play (client vs. agent) | 15-30 min | Skill practice |
| 案例研讨 | Real case analysis | 20-40 min | Advanced agents |
| 异议攻关 | Objection busting focus | 10-15 min | Weak point training |
| 综合考核 | Full simulation exam | 30-60 min | Level assessment |
Real-time coaching during training:
Agent Response
│
▼
[Natural Language Understanding] → Extract key claims, tone, strategy
│
▼
[Evaluation Engine]
├─ Product knowledge accuracy ✓/✗
├─ Objection handling effectiveness (1-5)
├─ Compliance adherence ✓/✗
├─ Closing attempt timing (good/early/late/missing)
├─ Client empathy signals ✓/✗
└─ Product combination logic ✓/✗
│
▼
[Real-time Coaching Feedback]
├─ Immediate tip (if struggling): "💡 提示:可以先问客户目前的保障缺口..."
├─ Completion praise (if excellent): "🌟 完美!您已经很好地识别了客户需求"
└─ Post-question summary: "本轮得分 85/100。建议加强:竞品对比环节"
Training session flow:
1. 导入 (5%) → 介绍训练目标和产品背景
2. 暖场 (10%) → 快问快答热身,激活产品知识
3. 主体 (60%) → 情景对练:客户角色扮演 + 实时点评
4. 复盘 (20%) → AI给出详细反馈:优点/不足/改进建议
5. 行动 (5%) → 下次拜访的具体行动计划
6. Effect Assessment & Progress Tracking / 效果评估与进度追踪
Metrics tracked per session:
| Metric | Definition | Target |
|---|---|---|
| 产品知识得分 | 知识点正确率 | L1: ≥70%, L2: ≥80%, L3: ≥90% |
| 异议处理时效 | 从异议提出到满意回答的时间 | < 30秒 |
| 促成成功率 | 能否自然引入促成信号 | ≥ 1次有效尝试 |
| 话术合规率 | 合规敏感词使用正确性 | 100% |
| 方案完整性 | 保障覆盖广度 | ≥ 3个维度 |
Progress report structure:
## 📊 代理人张明 训练报告 - 2026-05-05
### 综合得分: ⭐⭐⭐⭐ (78/100)
| 维度 | 本次得分 | 较上次 | 目标 |
|------|---------|--------|------|
| 产品知识 | 82/100 | ↑5 | 80+ |
| 异议处理 | 71/100 | ↓3 | 75+ |
| 促成技巧 | 85/100 | ↑8 | 80+ |
| 合规话术 | 95/100 | →0 | 100 |
| 方案设计 | 72/100 | ↑12 | 75+ |
### 🔥 本次表现亮点
1. 养老规划方案逻辑清晰,能结合客户生命周期讲解
2. 合规话术使用规范,犹豫期/退保说明完整
### ⚠️ 需要加强
1. 健康险异议处理:回应"已有社保"时过于被动,应主动算账
2. 竞品对比:对中国平安主要产品线不够熟悉
### 📅 明日训练重点
- 产品:康健医疗保险(健康告知流程)
- 场景:竞品替换(平安福 vs. XX福)
- 时长:30分钟情景对练 + 10分钟快问快答
Workflow / 标准工作流程
Mode 1: Quick Start (已知产品 + 快速训练)
User: "帮我准备明天拜访客户B的训练,他是私企老板,对健康险感兴趣"
│
▼
[Step 1] 获取代理人信息 → 张明,L2,弱项:健康险异议处理
[Step 2] 识别拜访产品 → 康健医疗保险(目标:替换平安福)
[Step 3] 生成训练计划 → 午间30分钟:健康险异议处理对练
[Step 4] 开始陪练 → 情景对练:私企业主健康险需求挖掘
[Step 5] 实时反馈 → 异议处理评分:71/100,给出改进建议
[Step 6] 报告输出 → 训练报告 + 明日拜访话术优化建议
Mode 2: Product Document Upload (上传产品文档)
User: [上传 XX保险公司福享人生终身寿险 产品手册 PDF]
│
▼
[Step 1] 解析文档 → 提取产品结构、条款、卖点
[Step 2] 生成产品画像 → Structured JSON Profile
[Step 3] 生成问题库 → 65+道题目(5难度×8类别)
[Step 4] 与现有产品库合并 → 更新知识库
[Step 5] 等待选择 → "请选择训练模式:快问快答 / 情景对练 / 案例研讨"
Mode 3: Full Agent Assessment (全面能力评估)
User: "帮我评估代理人李华的综合能力,她入职8个月,主要卖重疾险"
│
▼
[Step 1] 建立代理人档案 → L1入门级,8个月,重疾险方向
[Step 2] 产品文档上传 → 重疾险产品手册
[Step 3] 综合考核 → 30题产品知识 + 5个情景对练
[Step 4] 生成能力雷达图 → 6维度能力可视化
[Step 5] 制定成长路径 → 90天训练计划
Input / Output Specifications / 输入输出规范
Input
| Input Type | Description | Example |
|---|---|---|
| 代理人档案 | JSON/文本描述 | 姓名、级别、工龄、业绩、弱项 |
| 产品文档 | PDF/Word/TXT/图片 | 保险产品手册、条款、计划书 |
| 当日行程 | 文本/日历 | 09:00晨会 / 10:30拜访客户A |
| 训练指令 | 自然语言 | "帮我准备健康险的陪练" |
| 客户信息 | 文本描述 | "45岁私企老板,年收入200万" |
Output
| Output Type | Description |
|---|---|
| 产品画像JSON | 结构化产品信息 |
| 问题库 | 65+道分类分级题目 |
| 训练计划 | 分钟级个性化日程 |
| 陪练对话 | 实时AI角色扮演 |
| 评估报告 | 评分 + 改进建议 + 雷达图 |
| 成长路径 | 30/60/90天训练建议 |
Integration Notes / 集成说明
Data privacy:
- All agent and client data remains local / within the company's system
- No sensitive PII should be included in training documents
- Comply with China CBIRC insurance sales compliance regulations
Lianxi with other Skills:
insurance-bidding-pro: Use product analysis for bidding scenariosinsurance-private-domain-ops: Link training completion to customer follow-upinsurance-claims-intelligence: Train agents on claim processes for better client communication
Disclaimer / 免责声明
⚠️ Training is advisory only. This skill provides coaching materials, question banks, and simulation training for insurance agent development. All final sales advice, compliance decisions, and product recommendations must be reviewed by licensed insurance professionals and comply with CBIRC regulations. Model answers represent reference best practices, not guaranteed outcomes.