🎯 Objective

By the end of this assignment, you will be able to:

• Define a useful task that could be handled by an AI agent

• Set clear rules for when the agent should act independently and when it should involve a human

• Choose a tool or platform to implement your agent (Zapier, Make.com, or VectorShift)

• Describe your agent’s workflow using a simple 3–4 step structure

📝 Step-by-Step Instructions

Step 1: Identify a Use Case for an AI Agent

Think of a task in your role or department that involves some decision-making but follows consistent logic.

💡 Examples:

• Responding to simple support tickets

• Prioritizing incoming requests

• Drafting candidate screening summaries

• Sorting new leads into CRM pipelines
  
  

My agent will help with:
→ __________________________________________

Step 2: Set Guardrails (When to Act, When to Ask)

Decide when your agent can operate alone, and when it should ask for human approval.

Scenario                                         Can the agent act alone? (Yes/No)                What should it do if it needs help?

Standard request

Complex or sensitive case

Example:

• For password reset emails → ✅ act alone

• For billing disputes → ❌ ask support lead to review

Step 3: Outline Your Agent Workflow

Use this 4-step structure to define your workflow:

1. Trigger – What starts the process?

2. AI Action – What does the AI analyze or decide?

3. Decision Logic – If conditions are met, what happens next?

4. Output – Where does the result go?
   
   

My Agent Workflow:

1. Trigger: _____________________________________

2. AI Action: ___________________________________

3. Decision Logic: _______________________________

4. Output: _____________________________________
   
   

Step 4: Choose a Platform

Which platform would you use to build this AI agent?

🔘 Zapier + ChatGPT
🔘 Make.com
🔘 VectorShift
🔘 Not sure yet

Step 5: Optional – Write the AI Prompt

If your agent is using ChatGPT, what would you like it to do?

💡 Example:

“Summarize this support message in 3 bullet points. If it contains a refund request, flag it as high priority.”

My prompt would be:
→ __________________________________________________________

🤔 Reflection (Optional)

• What benefits would this agent bring to your team?

• What risks or limitations should you monitor?

• What would be the first small version (MVP) of this agent you could try?