product-roadshow

Product Roadshow Plan (Product Roadshow)

When the user needs to generate a product roadshow plan for a specific client, execute according to the following process.

Target Role (Target Role)

Role: Corporate client manager, product manager
Use scenario: Product promotion preparation before client visits, preparation for special marketing activities, response to client needs
Output purpose: Generate a personalized product roadshow plan to help client managers achieve effective marketing during visits
Decision level: Information assistance and marketing strategy guidance, does not involve credit approval decisions
Execution frequency: Execute once before each important client visit or special marketing activity

Data Sources (Data Sources)

Required data

Data item	Source	Acquisition method	Sensitivity level
Client basic information	In-bank client information system	API: /api/customer/basic	Internal
Current status of client product holdings	Product management system	API: /api/customer/products	Internal
Deposits and AUM data	Core banking system	API: /api/customer/deposits	Internal
Credit usage status	Credit management system	API: /api/customer/credit	Confidential
Industry market data	Industry research knowledge base/external data sources	Retrieval/online search	Public
Competitor bank interest rate market conditions	Market information base/client manager interviews	Retrieval/oral account	Public/Internal
Product spectrum and pricing	Product management system/pricing system	API: /api/products/catalog	Internal

Data masking rules

Client ID number: display the first 3 and last 4, replace the middle with *
Bank account number: display only the last 4 digits
Client sensitive financial data: use only in internal plans, must not be disclosed externally
Specific competitor bank interest rates: use market ranges, do not label specific bank names
Client trade secret information (such as core client lists): use "a certain client" instead in the plan

Degradation strategy

If product holding data is unavailable: mark "product holding status not verified", infer needs based on client industry characteristics
If competitor bank data is unavailable: mark "competitive analysis based on public market information", use LPR plus/minus point ranges
If industry market data is unavailable: mark "industry analysis based on client manager oral account", reduce analysis depth
If client financial data covers only 1 year: mark "insufficient data, trend analysis unavailable", perform only static analysis
If the client 360 system is unavailable: use the client background information output by the pre-visit-plan Skill
If the product spectrum is unavailable: annotate "Product recommendations are based on relationship manager experience," reducing recommendation accuracy.

Execution Workflow

Step 0: Data Confirmation and Validation

List the input parameters: customer name, roadshow scenario, key products (optional), competitive background (optional). Confirm the data time range and product spectrum version. Run scripts/validate_roadshow.py to check the completeness of the input parameters.

✅ Validation passed → proceed to Step 1
❌ Validation failed → return the missing list and ask the user to supplement it

📋 Data source:user_upload(customer information provided by the user)

Step 1: Obtain Core Customer Data

Query the customer 360-degree panoramic view to obtain:

Asset overview (AUM scale, average daily deposits, deposit structure)
Credit balance and usage status (used quota, remaining quota, guarantee method)
Contribution to intermediary business income (settlement fees, wealth management subscription fees, etc.)

The data acquisition step must not be skipped, and key figures must be extracted.

📋 Data source:system_api(in-house customer information system)

Step 2: Read the Customer's Existing Business Materials

List and prioritize reading the customer's business files:

Historical roadshow plans (the most recent 1-2 times)
Signed contract texts
Customer visit records (the most recent 3 times)
If the user specifies particular files, prioritize reading those files

📋 Data source:system_api(image archive system)

Step 3: Analyze the Customer's Current Product Holdings

Sort out the list of products the customer already holds:

Credit products (working capital loans, fixed asset loans, bills, trade finance)
Deposit and funding products (large-denomination certificates of deposit, structured deposits, wealth management)
Settlement and account products (online banking, payroll agency services, bill pools)
Identify blank products and low-penetration products

This step must be executed, and products the customer already holds must not be recommended.

📋 Data source:system_api(product management system)

Step 4: Industry Background Assessment

Understand the capital operation characteristics of the industry to which the customer belongs:

Industry capital turnover characteristics (seasonal patterns, settlement cycles)
Typical financing needs (working capital loans, fixed asset loans, supply chain financing)
Industry policy developments (in the past 3 months)

Consult references/product-catalog.md to obtain industry-appropriate product recommendations.

📋 Data source:reference(in-house industry knowledge base) or online search

Step 5: Competitive Landscape Assessment

Understand the existing business of major competing banks with this customer:

Distribution of competing banks (as stated by the customer or from historical records)
Current market interest rate conditions (LPR spread range)
Our bank's differentiated advantages (approval efficiency, exclusivity of service, product innovation)

The LPR spread range must be used, and the data source and date must be annotated.

📋 Data source:user_upload(as stated by the relationship manager) or market information database

Step 6: Diagnose Core Needs

Based on the above information, identify the 1-3 highest-priority financial need scenarios:

Operating cycle needs (seasonal funding gaps, investment and capacity expansion)
Liquidity needs (short-term turnover, accounts receivable financing)
Settlement efficiency needs (cross-border settlement, group fund concentration)
Asset preservation needs (use of idle funds, foreign exchange risk hedging)

Do not skip needs diagnosis and directly recommend products; it must reflect "why this customer needs it".

📋 Executing entity:ai

Step 7: Product portfolio design

Based on the needs scenario, select the best-matching combination from the product spectrum:

No more than 3 core recommended products
No more than 2 secondary products
Avoid piling up product lists; focus is better than coverage

Refer to references/product-catalog.md to obtain the complete product spectrum and applicable scenarios.

📋 Executing entity:ai

Step 8: Generate a personalized plan

Following the output template structure, generate a customized roadshow plan for this customer:

Needs diagnosis (reflecting the customer's industry characteristics)
Product portfolio recommendation (including applicable scenarios and quantitative data)
Competitor comparison analysis (based on real market conditions)
Return calculations (client side and bank side)
Marketing script (personalized, including cases)
Follow-up plan (executable and verifiable)

All return calculations must indicate the calculation basis; vague expressions such as "substantially increase" or "significantly save" must not be used.

📋 Executing entity:ai 📋 Confirmation mechanism:inform(notify the relationship manager for confirmation after generation)

Step 9: Plan formatting and validation

Format the output according to the assets/roadshow-plan-template.md template. Run scripts/validate_roadshow.py to validate plan completeness:

Whether the needs diagnosis section exists
Whether the number of products complies with the limits (core ≤3, secondary ≤2)
Whether return calculations indicate assumptions
Whether risk warnings are included (if there are wealth management/insurance products)
Whether a disclaimer is included
✅ Validation passed → output the final plan
⚠️ Partially abnormal → mark missing items and then output
❌ Validation failed → return the repair checklist and regenerate

📋 Executing entity:ai

Compliance constraints (Constraints)

Prohibition of return promises: Under no circumstances may promissory expressions such as "expected to obtain X ten thousand in deposits" or "guaranteed successful marketing" be given.
Prohibition of data guessing: Missing data = ask the user for it or mark "data missing" in the plan; it is strictly prohibited to replace actual data with industry averages.
Needs diagnosis priority: Product recommendations must be based on the customer's real needs; piling up product lists is prohibited.
Competitive sensitivity: Competitor comparisons must be based on actual market condition ranges; fabricating competitor data or disparaging competing banks is not allowed.
Compliance boundaries: Introductions to wealth management/insurance products must include risk warnings and must not promise principal protection or returns.
Personalized customization: The plan must reflect the customer's industry characteristics and business conditions; using generic template wording is prohibited.
Prohibition of unauthorized advice: This Skill only provides marketing strategy suggestions and does not involve credit approval decisions or limit approvals.
Prohibition of skipping steps: Do not skip needs diagnosis (Step 6) and directly recommend products, and do not ignore the current product holdings status (Step 3).
Mandatory red-line enforcement: If any red line (R1-R3) is triggered, generation must stop immediately and the user must be prompted.

Financial Compliance Red Lines (One-Vote Veto)

In the following situations, generation must be stopped immediately and the user prompted:

R1: The user requests a commitment to principal protection and guaranteed returns (wealth management products)
R2: The user requests fabricating data on competitors' disadvantages
R3: The user requests concealing product risks or fees

Audit Trail

After each product roadshow proposal generation is completed, generate an audit log audit/{客户简称}_{日期}_audit.json:

{
  "skill_name": "product-roadshow",
  "skill_version": "2.0.0",
  "execution_time": "2026-05-05T10:30:00+08:00",
  "input_params": {
    "customer_name": "XX企业",
    "roadshow_type": "全量产品推介/专项融资方案/存款提升方案/综合金融服务",
    "focus_products": ["流贷", "票据池"]
  },
  "operator": "客户经理姓名(工号:XXX)",
  "steps": [
    {
      "step": "获取客户核心数据",
      "executor": "ai",
      "data_source": {"type": "system_api", "system": "客户信息系统"},
      "result": "pass"
    },
    {
      "step": "分析客户产品持有现状",
      "executor": "ai",
      "data_source": {"type": "system_api", "system": "产品管理系统"},
      "result": "pass"
    },
    {
      "step": "生成个性化方案",
      "executor": "ai",
      "data_source": {"type": "context"},
      "confirmation": {"type": "inform", "notified_to": "客户经理姓名", "notified_at": "2026-05-05T10:35:00+08:00"},
      "result": "pass"
    }
  ],
  "warnings": ["部分行业数据基于公开信息"],
  "references_used": ["references/product-catalog.md"]
}

Audit log retention period: at least 3 years.

Product Lineage Reference

The detailed product lineage has been moved to references/product-catalog.md, including:

Credit products (working capital loans, fixed-asset loans, supply chain financing, bills, trade financing)
Deposits and cash management (large-denomination certificates of deposit, structured deposits, wealth management, cash management)
Settlement and account services (online banking, API direct connection, bill pool, payroll agency)
Investment and financing advisory (bond underwriting, syndicated loans, M&A loans)

Output Format

Use the assets/roadshow-plan-template.md template. The proposal must include the following sections:

Section	Data Type	Description
1. Customer Needs Diagnosis	Text + Table	Business characteristics profile, needs priority assessment
2. Recommended Product Portfolio	Table	Core recommended products (≤3), product portfolio synergy logic
3. Competitor Comparison Analysis	Table (5 dimensions)	5-dimension comparison table, core differentiation advantages
4. Return Calculation	Table + Text	Client-side value, bank-side value (calculation basis must be indicated)
5. Marketing Script	Text	Opening introduction, needs confirmation, product introduction, objection handling, closing for commitment
6. Follow-up Plan	Table	Action items at 3 time points (including verifiable outputs)

All data annotations: data source + data date + whether it is the latest data.

Downstream Compatibility: This output can be parsed and used by the submit-credit-application Skill. Key fields include:

Customer name (for credit application)
Recommended product portfolio (for credit product selection)
Return calculation (for credit limit reference)

Disclaimer: At the end of the proposal, the shared/disclaimer-template.md template must be cited, including necessary statements such as "This proposal does not constitute investment advice and is for reference only."

Gotchas

#1: Product stacking instead of needs-driven

Symptom: The proposal recommends 8-10 products, and after reading it the customer does not know what the focus is
Cause: The principle of prioritizing needs diagnosis was not strictly implemented, and products were selected directly from the product lineage
Solution: No more than 3 core recommended products, no more than 2 secondary products, and each product must explain "why this customer needs it"

#2: Competitor comparison lacks data support

Symptom: Writing "our bank's interest rate is better" but without specific numbers, so the customer does not trust it
Cause: Did not check actual market conditions, or used vague expressions
Solution:Competitor comparison must use the LPR spread range, and the data source and date must be noted

#3:Return calculation assumptions are not transparent

Symptoms:Writing "annual cost savings of XX ten-thousand yuan" without explaining the calculation basis
Cause:Calculation assumptions are not noted, so the customer cannot verify them
Solution:All return calculations must be labeled "Calculation basis: [customer data + assumptions]"

#4:Scripts are too generic and lack specificity

Symptoms:The script applies to any manufacturing customer and does not reflect this customer's characteristics
Cause:The customer's industry name, specific figures, and real cases are not embedded
Solution:The script must include "the customer's industry + recent developments + specific figures"

#5:Ignoring the current product holdings

Symptoms:Products the customer already holds were recommended
Cause:The holdings list in the product management system was not checked
Solution:Step 3 must be executed to identify blank products and low-penetration products

#6:Data timeliness is not noted

Symptoms:The plan uses "current market interest rate" but does not note the data date
Cause:The data labeling specifications were not strictly implemented
Solution:All data must be labeled "data source + data date + whether it is the latest data"

Examples

Example 1: Financing and expansion roadshow for a manufacturing customer

User input:

请为XX制造有限公司生成产品路演方案。客户成立于2018年,现有授信3000万流贷即将到期,计划扩建新厂房,需要5000万固贷。主要竞争行是工行和建行。

Skill execution process:

Obtain core customer data → AUM 20 million, working capital loans used 28 million, average daily deposits 15 million
Read historical materials → Last visit record shows customer orders increased by 30%
Current product holdings → Holds working capital loans, online banking, and payroll agency services; does not hold fixed-asset loans or bill pools
Industry background assessment → Metal products industry, mature stage, capital-intensive
Competitive landscape → ICBC working capital loan rate LPR+50BP, CCB LPR+45BP
Diagnose core needs → Fixed-asset loan (plant expansion), working capital loan renewal, bill pool (upstream and downstream settlement)
Product combination design → Core: fixed-asset loan 50 million + working capital loan renewal 30 million; secondary: bill pool
Generate plan → Plan type "special financing plan"

Output points:

Core needs: fixed-asset loan 50 million (plant expansion), working capital loan renewal 30 million
Product combination: use the fixed-asset loan as the main line to drive working capital loan renewal and the bill pool
Competitor comparison: our bank has fast approval turnaround (15 days vs ICBC 25 days), interest rate LPR+40BP
Return calculation: the customer saves 150,000 yuan in interest per year, and the bank earns 2 million yuan in interest income per year

Example 2: Comprehensive financial services roadshow for a trading company

User input:

请为YY贸易有限公司生成综合金融服务路演方案。客户主要做进出口贸易,年流水5亿,目前有中行和招行业务。

Skill execution process:

Obtain core customer data → AUM 50 million, no credit line, deposits mainly demand deposits
Current product holdings → Only holds a basic account; does not hold online banking, wealth management, or trade finance
Industry background assessment → Wholesale industry, frequent cross-border settlement, exchange-rate sensitive
Competitive landscape → Bank of China is mainly cross-border settlement, China Merchants Bank is mainly wealth management
Diagnose core needs → Cross-border settlement, foreign exchange hedging, wealth management appreciation, online banking
Product portfolio design → Core: letters of credit + foreign exchange hedging + structured deposits; Secondary: online banking API direct connection
Generate proposal → Proposal type "Comprehensive Financial Services"

Output highlights:

Core needs: cross-border settlement efficiency, exchange rate risk control, idle fund appreciation
Product portfolio: with letters of credit + foreign exchange hedging as the main line, driving structured deposits and online banking
Competitor comparison: our bank's cross-border settlement fee is 0.5‰ lower, and the foreign exchange hedging spread is 10BP less
Return calculation: the client saves 250,000 in settlement costs annually, while the bank contributes 500,000 in intermediary income

Non-functional scope (Out of Scope)

This Skill does not handle credit approval decisions, and only provides marketing strategy recommendations
This Skill does not generate post-loan management reports (please use the Post-Loan Management Skill)
This Skill does not handle risk investigation or risk warning
This Skill does not handle product promotion for personal credit/retail business
This Skill does not directly execute product signing or transaction operations
If the user requests the above content, clearly inform them and recommend a suitable Skill or contact the corresponding department

Differentiated proposal strategy (automatically adjusted according to the type of need)

Customer need type	Recommended product focus	Core entry point of the pitch
Financing expansion type	Fixed asset loans + project loans	Matching of repayment sources, flexibility of installment repayment
Tight working capital type	Revolving credit + accounts receivable financing	Borrow and repay at any time to save costs, fast disbursement
Deposit appreciation type	Structured deposits + large-denomination certificates of deposit	Principal-protected return comparison, deposit term matching
Supply chain settlement type	Bill pool + supply chain financing	Upstream and downstream integration, bill discounting efficiency
Cross-border business type	Letters of credit + cross-border settlement + foreign exchange hedging	Exchange rate risk control, cross-border cost optimization
Comprehensive financial services type	Payroll agency + online banking + wealth management	Overall migration value, one-stop service

Input parameter description

Parameter	Description	Required
Customer name	Full enterprise name or in-bank customer ID	Required
Roadshow scenario	Financing needs/deposit growth/comprehensive finance/specific product promotion	Recommended to fill in
Key products	Products that the user specifies they hope to focus on promoting	Optional
Competitive background	Known competitor bank situation or the customer's current main bank	Optional
Specified files	Customer files or historical proposal files checked by the user	Optional, read with priority

About this skill

Target Role (Target Role)

Data Sources (Data Sources)

Required data

Data masking rules

Degradation strategy

Execution Workflow

Step 0: Data Confirmation and Validation

Step 1: Obtain Core Customer Data

Step 2: Read the Customer's Existing Business Materials

Step 3: Analyze the Customer's Current Product Holdings

Step 4: Industry Background Assessment

Step 5: Competitive Landscape Assessment

Step 6: Diagnose Core Needs

Step 7: Product portfolio design

Step 8: Generate a personalized plan

Step 9: Plan formatting and validation

Compliance constraints (Constraints)

Financial Compliance Red Lines (One-Vote Veto)

Audit Trail

Product Lineage Reference

Output Format

Gotchas

#1: Product stacking instead of needs-driven

#2: Competitor comparison lacks data support

#3:Return calculation assumptions are not transparent

#4:Scripts are too generic and lack specificity

#5:Ignoring the current product holdings

#6:Data timeliness is not noted

Examples

Example 1: Financing and expansion roadshow for a manufacturing customer

Example 2: Comprehensive financial services roadshow for a trading company

Non-functional scope (Out of Scope)

Differentiated proposal strategy (automatically adjusted according to the type of need)

Input parameter description