Part 01 prepared and tested the four specialist AI agents. Part 02 activates those agents inside AI Configuration, connects customer intents to the correct specialist, preserves recent context, and defines human-assignment behavior. Part 03 will validate the completed system through the practical WhatsApp journey.
A Practical Guide to Building Specialized AI Agents, Conversation Routing, Contextual Memory, and Human Handoff
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 02 of 03: Connect, Route, and Control the Agents
Step 1: How Do You Open AI Configuration?
Go to Chatbot Manager > Bots/Accounts > select Promotute > AI > AI Configuration. Turn on AI Agent so the configured specialists can start responding with their training data.
Step 2: Which AI Agents Should Be Active?
Under Active Agents, select the four specialists prepared for the bot: Sales Agent SG, Order Agent SG, Lead Qualification Agent SG, and Support Agent SG. Only selected Active Agents can participate in routing and agent-to-agent transfers.
Step 3: How Do You Configure Agent Routing?
Use Agent Routing to describe which type of customer message should start with each specialist. Click Add Rule and create one rule for each agent.
AI Configuration showing four Active Agents and four routing rules
Step 4: Which AI Response Mode Should You Use?
For this setup, keep AI Agent for All Queries enabled so incoming customer messages are handled through the active-agent routing rules. Keep AI as a Fallback Only disabled when AI Agents are the primary response layer.
Step 5: How Does Contextual Memory Work?
Enable Contextual Memory so recent conversation details remain available when a customer moves between specialists. In this example, the Memory Window is set to 10 conversations. Run a contextual memory test by sharing a detail with one agent, transferring the conversation, and confirming that the next agent can use the relevant recent context without asking the customer to repeat it.
Step 6: How Does Human Assignment Work?
Enable Assign to Team or Department when a conversation may need human support. Select the appropriate team and write a short customer-facing Assignment Message. This creates a real-time agent assist path: AI handles routine work, while a person can take over when the request needs human judgment or the customer explicitly asks for support.
Step 7: Which Misc. Settings Should You Enable?
Use these controls to refine the live conversation experience. Enable business-off-hour assignment protection when needed and enable the typing indicator for a more natural WhatsApp experience.
Step 8: How Do Behavior & Restrictions Work?
Add Restricted Topics only for subjects the AI should not handle. Define a clear response that is returned when a restricted topic is detected.
Step 9: How Do Business Hours Affect Human Handoff?
Set the correct time zone and working hours for the human team. In this example, the AI Configuration uses Asia/Dhaka with 09:00 AM to 06:00 PM. Human assignment should follow the actual support schedule used by the business.
Step 10: How Should You Configure Advanced Settings?
Set the AI reply delay and reasoning level. In this example, Delay for AI Reply is 0 seconds and AI Reasoning Level is Medium.
Response Mode, Context & Memory, Assignment, Misc. Settings, Restrictions, Business Hours, and Advanced Settings
Step 11: How Do You Test the Multi-Agent Configuration?
Click Save Settings after reviewing the configuration. Then test different customer messages to confirm that each intent reaches the correct specialist.
After basic routing works, test agent-to-agent transfers, reservation flows, order actions, and human handoff separately.
Continue to Part 03: See the Multi-Agent System in Action
The specialist agents are now prepared, activated, and connected through routing rules. BotSailor knows which agent should handle each type of customer request and how conversations can move between specialists.
Part 03 brings everything together in a complete WhatsApp journey covering Sales, Lead Qualification, the reservation flow, Order, Support, and human handoff.
Part 02: Configure the System → Part 03: (See the Multi-Agent System in Action): Test the Complete Customer Journey
