# How to Build Multiple Instagram AI Agents in ZazzyAgent As your automation becomes more advanced, one AI Agent may become responsible for too many things. Imagine asking one agent to handle: * New leads * Product questions * Pricing * Existing customers * Complaints * Support * Sales * Human handoff That's where multiple specialized AI Agents become useful. Instead of one general-purpose AI, create several agents with clearly defined responsibilities. ## Example AI team A simple setup can use three agents: ### Lead Qualification Agent Handles new enquiries and collects lead information. ### Sales Agent Handles product, pricing and purchase conversations. ### Support Agent Handles existing customers and support requests. The current multi-agent Instagram workflow supports specialized agents, individual prompts/knowledge/actions and routing between agents. ## Why use multiple agents? Specialization makes instructions simpler. A Lead Agent doesn't need to know everything about support. A Support Agent doesn't need to qualify every new lead. A Sales Agent doesn't need to manage every conversation. Each agent can have: * Its own System Prompt * Its own Knowledge * Its own Actions * Its own routing logic ## Agent 1: Lead Qualification Its job: > Identify and qualify new Instagram enquiries. It may collect: * Name * Email * Requirement * Budget * Timeline Then save the information and transfer the lead. ## Agent 2: Sales Its job: > Help qualified leads make a purchase decision. It may answer: * Product questions * Pricing questions * Plan comparisons * Feature questions Once the customer is ready to buy, it can hand the conversation to sales staff. ## Agent 3: Support Its job: > Help existing customers. It may handle: * Common support questions * Policies * Basic troubleshooting * Human escalation ## Create the agents Create each agent separately in the AI Agent area. Use clear names. Examples: > Instagram Lead Agent > Instagram Sales Agent > Instagram Support Agent Give each a specific description. ## Give each agent only the Knowledge it needs ### Lead Agent General business FAQs + qualification information. ### Sales Agent Products + pricing + sales FAQs. ### Support Agent Support FAQs + policies + troubleshooting information. This prevents unnecessary overlap. ## Give each agent only the actions it needs For example: ### Lead Agent * Save Custom Field * Add Label * Transfer Agent ### Sales Agent * Save Custom Field * Assign Human * Transfer Agent ### Support Agent * Assign Human * Mark Solved * Skip Reply The exact actions depend on your use case. ## Configure routing Once the agents are created, open **Agent Configuration**. Enable the agents for your Instagram account. Then define which conversations should go to which agent. A simple model is: ```text New enquiry ↓ Lead Agent Pricing / product questions ↓ Sales Agent Existing customer / support ↓ Support Agent ``` ## Think about intent, not only keywords A customer might say: > I want to know how much your service costs. That's a sales conversation. Another customer might say: > I bought this last week and need help. That's support. The routing system should be designed around these kinds of conversation signals. ## Transfer from one agent to another A lead can naturally move through the system. Example: ```text Customer ↓ Lead Agent ↓ Qualification ↓ Transfer ↓ Sales Agent ↓ Purchase ↓ Human ``` The customer doesn't need to understand the internal architecture. ## Don't create routing loops Avoid: ```text Lead → Sales → Lead → Sales ``` The customer should not be passed back and forth unnecessarily. Keep transfer conditions clear. ## Example conversation Customer: > Hi, I'm looking for WhatsApp automation for my company. **Lead Agent** > Great. What are you hoping to automate? Customer provides details. Lead Agent: > Thanks. What is your approximate monthly message volume? After qualification: **Transfer → Sales Agent** Customer: > Which plan is best for me? Sales Agent answers using its Knowledge. Customer: > Great. I want to get started today. **Assign → Human Sales Team** This is a clean progression. ## What if a customer jumps directly to support? That's fine. A customer might start: > My order hasn't arrived. The routing system can send them directly to Support rather than forcing them through Lead Qualification. ## Keep each agent's prompt narrow A good prompt says: > Your job is to qualify new Instagram leads. A weak prompt says: > You are the company's AI assistant. Answer everything. The narrower the job, the easier the system is to manage. ## Test routing separately from the agents You need to test: ### Agent response Can the agent answer correctly? ### Routing Did the correct agent receive the conversation? ### Transfer Did the next agent receive it correctly? ### Human handoff Did the conversation reach the correct team? These are separate tests. ## Test these scenarios **Scenario 1** > Hi Expected: Lead Agent. **Scenario 2** > I'm interested in your service and want a quote. Expected: Lead Agent. **Scenario 3** > What's the price difference between your plans? Expected: Sales Agent. **Scenario 4** > I bought this last week and need help. Expected: Support Agent. **Scenario 5** > I want to speak to someone. Expected: Human escalation according to your rules. ## A complete Instagram AI team ```text Instagram ↓ Agent Routing ↙ ↓ ↘ Lead Agent Sales Agent Support Agent ↓ ↓ ↓ Qualify Sell Support ↓ ↓ ↓ └──────→ Human ←─────────┘ ``` This approach lets your Instagram account behave more like a real customer-service team. ## When should you stay with one agent? Use one agent when: * Your business is small. * Conversations are simple. * There are few use cases. * One prompt can handle everything reliably. Add multiple agents when: * Different teams handle different conversations. * The prompt is becoming very complicated. * Routing needs to change based on customer intent. * Sales and support require different Knowledge. ## The most important principle Don't create multiple agents just because you can. Create them when **specialization makes the customer experience better and the configuration easier to control**. A good multi-agent Instagram setup should feel like one intelligent business conversation to the customer, even though several specialized agents may work behind the scenes.