Skip to content

Customer Chat (AI Conversations) Documentation

Menu Location: Customers > Last Cust Viewed > Chat OR AI > Conversations

Access Level: Customer Service and above

Last Updated: 2026-03-01

Overview

The Customer Chat page displays AI-powered conversations about a specific customer, allowing you to ask natural language questions about their account, orders, preferences, and history. The AI assistant analyzes the customer's complete data and provides insights, summaries, and recommendations.

Primary Functions:

  • Ask questions about customer account in natural language
  • Get AI-generated summaries of customer history
  • Receive recommendations for customer service actions
  • Analyze customer behavior patterns
  • Generate personalized communication suggestions
  • Troubleshoot customer issues with AI assistance

Page Layout

Header Section

  • Customer Name - Displays customer being discussed
  • Customer ID - Quick reference
  • Conversation ID - Unique ID for this chat session
  • New Conversation Button - Start fresh chat about this customer
  • Return to Customer Button - Back to customer detail page

Chat Interface

  • Message History - Scrollable conversation display
  • Input Box - Type your questions here
  • Send Button - Submit your question to AI
  • Example Questions - Suggested queries to get started
  • Conversation Tools - Save, export, share conversation

AI Response Area

Displays AI-generated responses with:

  • Text Answers - Natural language explanations
  • Data Tables - Formatted order history, products, dates
  • Recommendations - Suggested actions
  • Source Links - Links to orders, products, or data mentioned
  • Confidence Indicators - How certain the AI is about answer

How to Use Customer Chat

Starting a Conversation

First Time:

  1. Open customer detail page
  2. Click "Chat" button in header
  3. Chat opens with greeting and customer summary
  4. Review AI's initial analysis
  5. Ask follow-up questions

Returning to Chat:

  • Chat history is saved
  • Previous questions and answers remain
  • Continue where you left off
  • Or start new conversation for fresh context

Asking Questions

Question Format:

  • Use natural, conversational language
  • Be specific about what you want to know
  • Ask one question at a time for clarity
  • Follow up with clarifying questions

Example Questions:

  • "What are this customer's top 3 favorite products?"
  • "Has this customer had any delivery issues?"
  • "When did they last order chicken?"
  • "Why did they cancel their subscription?"
  • "What's their average order value?"
  • "Do they typically add extra items?"
  • "Have they ever complained about quality?"
  • "What day of the week do they usually receive delivery?"

Understanding AI Responses

Response Format:

  • AI provides direct answer first
  • Supporting data shown below
  • Links to relevant orders or products
  • Recommendations or suggestions
  • Sources cited

Confidence Levels:

  • High confidence: AI is certain based on clear data
  • Medium confidence: AI found patterns but with some ambiguity
  • Low confidence: Limited data or unclear patterns

Example Conversations

Example 1: Understanding Customer Preferences

You: "What are this customer's favorite products?"

AI Response: 

Based on order history over the past 6 months, this customer's
top favorite products are:

1. Organic Apples (ordered 18 times, never substituted out)
2. Grass-Fed Ground Beef (ordered 16 times)
3. Organic Mixed Greens (ordered 15 times)

They consistently add these items to their orders and have never
made substitutions away from them. This suggests strong
preference for these products.

Recommendation: Ensure these items are always available for their
orders to maintain satisfaction.

Example 2: Investigating Issues

You: "Has this customer reported any problems recently?"

AI Response: 

Yes, this customer has reported two issues in the past 30 days:

1. March 10: Missing item - Chicken breasts were left out of delivery
   - Action: $15 credit applied
   - Note: Driver error confirmed

2. February 28: Quality concern - Strawberries arrived damaged
   - Action: Replacement sent with next order
   - Note: Packaging issue addressed

Recent notes indicate customer satisfaction improved after prompt
resolution of both issues.

Recommendation: Add "Quality Check" flag for next few orders to
ensure extra care with this customer's deliveries.

Example 3: Win-Back Strategy

You: "Why did this customer cancel?"

AI Response: 

Customer cancelled on February 15, 2026. Cancellation reason
selected: "Too Expensive"

Analysis of their account:

- Never used discount codes (despite 5 being sent)
- Average order value: $78 (higher than typical $65)
- Frequently added premium items
- Started reducing order frequency 6 weeks before cancellation
- Reduced addons starting 3 weeks before cancellation

Pattern suggests: Price sensitivity developed over time, possibly
due to budget constraints rather than dissatisfaction with service.

Win-back recommendation:

1. Send targeted discount offer (15-20% off)
2. Highlight lower-cost box options
3. Mention ability to skip weeks to manage budget
4. Reference their favorite products to remind them of value

Common Use Cases

Use Case 1: Quick Customer Summary Before Call

Goal: Get up to speed on customer before phone conversation

Steps:

  1. Open customer chat
  2. Ask: "Give me a quick summary of this customer"
  3. Review AI's summary of key facts
  4. Ask follow-up questions as needed
  5. Proceed with informed phone call

What You Get:

  • How long they've been a customer
  • Order frequency and patterns
  • Recent issues or concerns
  • Favorite products
  • Communication preferences
  • Payment history

Use Case 2: Product Recommendation

Goal: Suggest products customer might like

Steps:

  1. Open customer chat
  2. Ask: "What products should I recommend to this customer?"
  3. Review AI's analysis of preferences
  4. Get specific product suggestions
  5. Use recommendations in communication

AI Considers:

  • Products frequently ordered
  • Products never ordered but similar to favorites
  • Seasonal patterns
  • Dietary preferences
  • Price sensitivity

Use Case 3: Troubleshooting Subscription Changes

Goal: Understand why customer modified subscription

Steps:

  1. Open customer chat
  2. Ask: "What changes has this customer made to their subscription?"
  3. Review timeline of modifications
  4. Ask: "Why do you think they made these changes?"
  5. Get AI's pattern analysis
  6. Use insights for retention strategy

Use Case 4: Personalize Marketing

Goal: Craft personalized promotional message

Steps:

  1. Open customer chat
  2. Ask: "What would motivate this customer to add more items?"
  3. Review AI's analysis of purchase drivers
  4. Ask: "Draft a promotional message for this customer"
  5. Get AI-generated personalized message
  6. Edit and send to customer

Use Case 5: Delivery Optimization

Goal: Improve delivery experience for customer

Steps:

  1. Open customer chat
  2. Ask: "Has this customer had any delivery issues?"
  3. Review delivery history
  4. Ask: "What can we do to improve their delivery experience?"
  5. Get AI recommendations
  6. Implement suggestions

AI Chat Capabilities

What AI Can Do

Account Analysis:

  • Summarize customer account status
  • Calculate lifetime value and metrics
  • Identify patterns in ordering behavior
  • Compare to typical customer profiles
  • Predict churn risk

Order History:

  • List all orders with details
  • Find specific products ordered
  • Calculate frequency of orders
  • Identify seasonal patterns
  • Track substitution preferences

Customer Service:

  • Suggest resolutions for issues
  • Draft personalized communications
  • Recommend retention strategies
  • Identify upsell opportunities
  • Flag accounts needing attention

Insights & Predictions:

  • Predict customer preferences
  • Identify churn risk factors
  • Suggest product recommendations
  • Analyze spending patterns
  • Forecast future behavior

What AI Cannot Do

Limitations:

  • Cannot make actual changes to account (you must do that)
  • Cannot send emails or messages directly
  • Cannot process payments
  • Cannot access external systems (only customer's data in your system)
  • May misinterpret complex edge cases
  • Cannot read customer's mind (relies only on data)

Troubleshooting

AI Gives Incorrect Information

Symptoms:

  • Answer doesn't match actual data
  • Numbers seem off
  • Contradicts what you see in account

Solutions:

  1. Rephrase question more specifically
  2. Ask AI to cite its sources
  3. Cross-reference with actual customer page
  4. Report issue if consistently wrong
  5. Start new conversation (may fix context issues)

Common Causes:

  • Question was ambiguous
  • AI misunderstood context
  • Data sync delay
  • Edge case not handled well

AI Says "I Don't Have Enough Information"

Symptoms:

  • AI cannot answer question
  • Responds with uncertainty
  • Asks for clarification

What to Do:

  1. Rephrase question more specifically
  2. Break complex questions into smaller parts
  3. Check if customer actually has data for that question
  4. Ask related but simpler question first
  5. Manually review customer account

Common Reasons:

  • New customer with limited history
  • Question about data not tracked in system
  • Question too vague or complex
  • Feature not yet implemented

Chat Won't Load or Times Out

Symptoms:

  • Chat interface doesn't open
  • Responses take too long
  • Page freezes

Solutions:

  1. Refresh page and try again
  2. Check internet connection
  3. Try different browser
  4. Clear browser cache
  5. Contact support if persistent
  • Customer Detail - Main customer account page
  • Customer Email History - Email communication log
  • Customer Log History - Account activity audit trail
  • AI Overview - System-wide AI features
  • Customer Chat View All - Browse all customer conversations

Permissions & Access

Required Access Level: Customer Service or higher

Access Level Capabilities:

  • Customer Service: Chat about assigned customers
  • Manager: Chat about all customers, save conversations
  • Administrator: All Manager + manage AI settings
  • Kiva Admin: All features + technical AI tools

Best Practices

Effective Questions

  1. Be specific and clear
  2. Ask one question at a time
  3. Use follow-up questions to dig deeper
  4. Rephrase if answer isn't helpful
  5. Verify important information manually

Using AI Insights

  1. Use AI as a starting point, not final answer
  2. Cross-reference important data
  3. Apply human judgment to recommendations
  4. Document AI insights that prove useful
  5. Share effective prompts with team

Privacy & Ethics

  1. Only use chat for legitimate business purposes
  2. Don't share AI conversations inappropriately
  3. Remember AI sees all customer data
  4. Be thoughtful about questions you ask
  5. Follow company AI usage policies

Quick Reference Card

Task Action/Location
Open customer chat Click "Chat" button in customer detail header
Ask question Type in input box and press Enter
Start new conversation Click "New Conversation" button
Get customer summary Ask "Give me a summary of this customer"
Find product preferences Ask "What are this customer's favorite products?"
Check delivery issues Ask "Has this customer had delivery problems?"
Get recommendations Ask "What should I recommend to this customer?"
Return to customer Click "Back to Customer" button
Save conversation Click "Save" button in chat tools
Export chat Click "Export" in conversation tools

FAQs

Is the AI reading my conversations with customers?

No, the AI only analyzes structured data in the customer's account (orders, products, notes, etc.). It doesn't read private conversations unless you paste them into the chat for analysis.

Can the AI make changes to the customer account?

No, the AI can only provide information and recommendations. You must make any actual changes through the normal admin interface.

How current is the information AI uses?

The AI uses real-time data from the customer's account. However, there may be a brief delay for very recent changes (usually seconds to minutes).

Can I trust the AI's recommendations?

AI recommendations are helpful starting points but should be verified and combined with human judgment. Always review the data AI cites and use your customer service experience.

What happens to my chat conversations?

Conversations are saved and can be reviewed later. They're private to you unless you explicitly share them. Old conversations may be archived after extended periods.

Can other admins see my chats?

Managers and administrators can view chat conversations for training and quality purposes. Standard customer service may only see their own chats.

Does using AI chat affect my performance metrics?

Using AI chat is encouraged and considered part of effective customer service. It doesn't negatively impact metrics when used appropriately.

What if the AI suggests something I disagree with?

Always use your judgment. The AI is a tool to assist you, not replace your decision-making. Document why you chose a different approach if you deviate from AI recommendations.

End of Documentation

For additional help, contact your system administrator or Kiva Logic support.