Self-Study Assignment 3.3: Think & Act 💡 From Friction to AI Agent Workflow

Learning Objectives Addressed

• Define Agentic AI and differentiate it from traditional automation and generative AI tools.
• Explain the core building blocks of agentic systems and their role in enabling autonomous workflows.
• Translate common workplace inefficiencies into potential agentic workflow opportunities.

Step 1 – Identify the Friction 

Think about a workflow you are personally involved in.

Identify one point of friction where delays, confusion, bottlenecks or repetitive manual work occur.

Examples:

• Approvals take too long.
• Customer queries get lost or delayed.
• Reports require repetitive manual effort.
• Information is scattered across multiple systems.
• Decisions require multiple handovers.
• Follow-ups are inconsistent.

Reflection Question

Where do delays, confusion or repeated manual work slow things down in your workflow?

Write down one specific friction point.

Step 2 – Define the Desired Outcome 

Now imagine that the friction disappears completely.

Define what success would look like.

Consider:

• Faster turnaround time
• Better accuracy
• Reduced manual effort
• Improved customer experience
• Better visibility and coordination
• Lower operational cost

Reflection Question

If this friction disappeared, what outcome would you see in speed, cost, quality or customer experience?

Be as specific as possible.

Step 3 – Design an AI Agent

Imagine you have access to an AI colleague that can support this workflow.

Design the role it would play.

Agent Role

What job would the agent perform?

Examples:

• Summariser
• Coordinator
• Recommender
• Checker
• Research Assistant
• Notifier
• Workflow Manager

Inputs

What information, data or documents would the agent need?

Actions

What tasks would the agent perform?

Human Handoffs

Where should the agent hand work back to a human?

Reflection Question

If you had an AI colleague who never slept, what specific responsibility would you assign to remove this friction?

Step 4 – Personal Action Plan 

Review your proposed workflow.

Consider:

• What benefits would this create?
• What risks or limitations would need oversight?
• What decisions should remain with humans?

Finally, identify one practical experiment you could run within the next 30 days.

Reflection Questions

1. What is the biggest benefit of your proposed AI agent?
2. What human oversight would still be required?
3. What is one small pilot or experiment you could run to test this idea?

Estimated Duration: 30 minutes

Note: This is a practice activity and will not be graded or impact your programme completion. However, we strongly encourage participation for a more holistic learning experience.