Quick Answer
Part 01prepared the specialist AI Agents, and Part 02connected them through AI Configuration and routing. This final partshows the outcome: a real WhatsApp journey where the customer moves between Sales, Lead Qualification, the reservation flow, Order, Support, and human assistance as their intent changes.
How to Configure Multi-Agent AI for WhatsApp Automation in BotSailor: Part-03
A Practical Guide to Building Specialized AI Agents, Conversation Routing, Contextual Memory, and Human Handoff
Naming Note
The suffix “SG” is used only for naming convenience and easy identification during this demonstration. It has no special technical meaning in BotSailor. Sales Agent SG, Order Agent SG, Lead Qualification Agent SG, and Support Agent SG are simply the names used in this guide.
Guide Scope
This guide reflects the BotSailor Multi-Agent AI interface and settings shown in the included screenshots and uses Promotute Restaurant as a practical demonstration. It explains how to configure and coordinate AI Agents in BotSailor; it does not teach how to create or train a new AI model from scratch.
Reader Journey:Part 01 Build the Foundation→Part 02 Connect, Route, and Control the Agents→Part 03 See the Multi-Agent System in Action
Part 03 of 03: See the Multi-Agent System in Action
Prepared Agents→Customer Conversation→Agent Transfer→Business Action→Support or Human Handoff
How Do the AI Agents Work Together in a WhatsApp Conversation?
The practical WhatsApp conversation below shows how a customer can move between specialist AI Agents, the reservation flow, order handling, support logic, and human handoff. It demonstrates how multi-agent AI can support autonomous customer service without allowing one agent to handle responsibilities outside its role.
Customer→Sales Agent SG→Lead Qualification Agent SG→Restaurant Table Reservation Flow→Order Agent SG→Support Agent SG→Human Support when available
Screenshot 01: Sales Agent SG - Greeting, Recommendation, and Sales Handoff
Description:The customer starts with a greeting and asks for spicy-food recommendations. The conversation moves into the restaurant sales journey, where Sales Agent SG recommends menu items and offers help with a reservation.
Screenshot 01
Conversation
Customer:Hi
Routing / AI:Hello! Welcome to Promotute. How can I assist you today?
Customer:Hi
Routing / AI:Hi there! How can I help you today?
Customer:I like spicy food. What do you recommend?
Routing / AI:We have a variety of spicy dishes. I can connect you with our Sales Agent for more specific recommendations and menu pricing. Would you like me to transfer you to a Sales Agent?
Customer:Yes
Sales Agent SG:I’d be happy to help you with spicy food recommendations and pricing. Here are some options from our menu:
• Chicken Burger - BDT 250
• Beef Burger - BDT 320
• Fried Chicken - BDT 220
Would you like to know more about any specific item, or shall I help you make a reservation?
Behind the scenes:Agent routing / transfer sends the customer to Sales Agent SG. Sales Agent SG answers from the restaurant knowledge source.
Screenshot 02: Sales Agent SG - Product Media
Description:The customer asks about the Chicken Burger. Sales Agent SG responds with product media and the item price, demonstrating the use of attached Media knowledge.
Screenshot 02
Conversation
Customer:Chicken Burger
Sales Agent SG:Here are some images of our delicious Chicken Burger!
Customer:Chicken Burger - BDT 250
Sales Agent SG:Here’s a look at our Chicken Burger!
BotSailor Media:Product image is displayed with the caption: Chicken Burger - BDT 250.
Behind the scenes:Sales Agent SG retrieves the relevant menu and media knowledge for the requested product.
Screenshot 03: Sales Agent SG - Product Detail and Purchase Intent
Description:The customer asks about the Beef Burger. Sales Agent SG provides pricing and product information, then checks whether the customer wants to order or reserve a table.
Screenshot 03
Conversation
Customer:Beef Burger
Sales Agent SG:Our Beef Burger is priced at BDT 320. It features a juicy grilled beef patty with melted cheddar cheese, fresh lettuce, tomato, and red onion on a toasted sesame bun. Comes with a side of golden fries!
Sales Agent SG:Would you like to place an order or reserve a table?
Customer:Yes
Sales Agent SG:Great! Would you like to place a delivery order or reserve a table?
Behind the scenes:Sales Agent SG continues the sales conversation and identifies the customer’s next intent.
Screenshot 04: Lead Qualification Agent SG - Group Booking Qualification
Description:The customer first appears ready to order, then changes the intent to a birthday dinner for around 20 people. Lead Qualification Agent SG collects the details needed before the reservation flow starts.
Screenshot 04
Conversation
Customer:Yes order
Sales Agent SG:Please provide delivery address, contact phone, quantity, and preferred delivery time for your Beef Burger order.
Customer:I'm actually planning a birthday dinner for around 20 people next Friday.
Lead Qualification Agent SG:What is the name for the booking?
Customer:Siddhartha
Lead Qualification Agent SG:Please provide your email address for the reservation.
Customer:siddharthaghosh07@gmail.com
Lead Qualification Agent SG:Thanks, Abir. I've saved your details: name Abir, email
siddharthaghosh07@gmail.com, booking: birthday dinner for around 20 people next Friday (Sept 18), may how you like to proceed. So, now I need some information for the completion of the Reservation flow time would you like to start the dinner?
Customer:8.00 pm
Lead Qualification Agent SG:Got it. Do you have a budget range in mind for the dinner?
Customer:around 20k
Reservation Flow:Welcome to our appointment booking system.
Behind the scenes:The lead agent can save the name, email, party size or requirement, preferred time, and budget into configured Custom Fields, apply a qualified-lead label, then move the conversation into the Restaurant Table Reservation flow.
Screenshot 05: Restaurant Table Reservation Flow - Date and Time Selection
Description:The structured reservation flow starts by showing the available date, then the available appointment times for the selected date, followed by the booking-time-duration step
Screenshot 05
Conversation
Reservation Flow:Select Appointment Date — Please choose your preferred date from the available options. Tap to select a date.
Customer Selection:Sep 18, 2026 — 20 slots available
Reservation Flow:Select Appointment Time — Please choose your preferred time from the available options. Date: Friday, Sep 18, 2026. Tap to select a time.
Customer Selection:More Available Times — Show next 1 of 1 remaining slots.
Reservation Flow:The system displays another Select Booking Time Duration message.
Behind the scenes:BotSailor Appointment Booking handles the structured date and time selection after the AI Agent triggers the reservation flow.
Screenshot 06: Restaurant Table Reservation Flow - Booking Confirmation
Description:The customer selects a time slot. BotSailor then generates the booking confirmation with the Booking ID, date, start time, end time, amount, and expiry time.
Screenshot 06
Conversation
Customer Selection:8:00 PM - 9:30 PM — 1.30 min • Asia/Dhaka
Reservation Flow:Confirm Booking
Reservation Flow:Hello Siddhartha, Thank you for booking your appointment with us! Please confirm your booking by clicking the Confirm button before it expires.
Reservation Flow:Booking ID: 1 | Date: 18th Sep 26 | Start Time: 08:00 pm | End Time: 09:30 pm | Confirmation Amount: USD 0 | Expiry Time: 15th Sep 26 21:30 +06:00
Reservation Flow:If you have any questions or need assistance, just reply to this conversation and we will be happy to assist you.
Action available:Confirm Booking
Behind the scenes:The Appointment flow creates the booking record and presents the final confirmation action.
Screenshot 07: Order Agent SG - Order Lookup and Cancellation Attempt
Description:Later, the customer asks about an existing order. Order Agent SG attempts a live lookup, does not invent the result when the order cannot be found, and then handles a cancellation request. The order API fails, so the agent reports the failure rather than falsely confirming cancellation.
Screenshot 07
Conversation
Customer:Where is my order ORD-1054?
Order Agent SG:I'm sorry, but I was unable to retrieve the details for order ORD-1054. It seems the system could not find this order at the moment. Could you please double-check the order number and try again?
Customer:Ok understand
Order Agent SG:Is there anything else I can help you with regarding your order or booking?
Customer:Actually, I want to cancel it.
Order Agent SG:Please provide the Order ID you wish to cancel.
Customer:I forgot Order ID
Order Agent SG:Please confirm if you want to cancel order ORD-1054.
Customer:yes
Order Agent SG:Sorry, I wasn't able to retrieve the order information — the system returned an error (order API not found). Could you please provide your order number so I can try another way to help you?
Behind the scenes:Order Agent SG uses the previous order context and attempts the configured order/API action. Because the API is unavailable, it safely reports the error instead of claiming the order was cancelled.
Screenshot 08: Support Agent SG - Human Support and Business Hours
Description:The customer asks for support and then explicitly requests a human. Support Agent SG explains the configured support hours. Because a human is not available at that time, the AI remains available and tells the customer when the support team will be available again.
Screenshot 08
Conversation
Customer:I want to talk to your support team
Support Agent SG:I understand you want to talk to the support team about your order cancellation.
Support Agent SG:Our support hours are:
Sunday: 11 AM - 3 PM;
Monday: 10 AM - 7 PM;
Tuesday - Wednesday: 10 AM - 1 PM;
Thursday - Saturday: 10 AM - 7 PM.
I'm here to help you 24/7 as an AI assistant. Could you please try to find your order number from your email confirmation or order history? Once you provide it, I can look into the cancellation for you right away.
Customer:I need Human Support
Support Agent SG:I understand you'd like to speak with a human. Here are the support hours (Asia/Dhaka timezone):
Sunday: 11 AM - 3 PM;
Monday: 10 AM - 7 PM;
Tuesday - Wednesday: 10 AM - 1 PM;
Thursday - Saturday: 10 AM - 7 PM.
They will be available again on Wednesday from 10 AM. In the meantime, I'm happy to help you with your order cancellation. If you can find your order number (check your email confirmation or order history), I can assist you right away!
Behind the scenes:Support Agent SG uses the human-handoff rule. When the configured human role is outside business hours, the AI remains available instead of pretending that a human has been assigned.
Complete Agent Journey
Customer→Sales Agent SG→Lead Qualification Agent SG→Restaurant Table Reservation Flow→Order Agent SG→Support Agent SG→Human Support when available
- Sales Agent SG: Menu recommendations, product prices, product media, sales intent, and reservation direction.
- Lead Qualification Agent SG: Collects booking name, email, group requirement, preferred time, budget, and qualifies the lead.
- Restaurant Table Reservation Flow: Handles available date/time selection and booking confirmation.
- Order Agent SG: Handles order lookup, order ID context, cancellation intent, and live API results or errors.
- Support Agent SG: Handles unresolved customer issues, support escalation, human requests, and business-hours logic.
Verification: How Do You Know the Multi-Agent AI Setup Is Working?
- All four prepared agents appear under Active Agents.
- Each routing rule sends the matching customer intent to the correct specialist.
- Sales Agent SG can answer menu questions and trigger the Restaurant Table Reservation flow.
- Lead Qualification Agent SG can collect the configured lead fields and pass qualified leads to Sales.
- Order Agent SG uses live data only when available and does not invent order or cancellation results.
- Support Agent SG can collect issue details, escalate to Restaurant Support, and respect business-hour behavior.
- Contextual memory is retained when needed across specialist agents.
- The saved configuration responds correctly during WhatsApp testing.
Common Troubleshooting
Key Takeaways for Multi-Agent AI Automation
- Use one specialist AI Agent for each clearly defined responsibility.
- Keep business facts in Knowledge Sources and behavior/action rules in the System Prompt.
- Select only prepared agents as Active Agents and write explicit routing rules for customer intents.
- Use Contextual Memory to preserve relevant recent information across specialist transfers.
- Use live APIs only when they are configured; never allow agents to invent order, payment, refund, or cancellation results.
- Keep human assignment and business hours available for issues that should not be handled by autonomous customer service alone.
Next Steps
Once the four agents, practical conversation flow, and routing configuration are working correctly, repeat the same tests with real business data. Keep Knowledge Sources current, use only actions backed by configured resources, review routing when new intents appear, and retest memory, API actions, reservation flows, support escalation, and human handoff whenever the workflow changes.
Prepare Agents→Test Individually→Review Target Conversation→Configure Routing→Save Settings→Test Full WhatsApp Journey→Refine with Real Business Data
Success
Your Multi-Agent AI setup is ready when the customer can move naturally between Sales, Lead Qualification, Order, Support, the reservation flow, and human support without the agents inventing live business information or handling responsibilities outside their role.
