Learning Outcomes Addressed: 

• Explain the evolution and business implications of Agentic AI
• Design and build no-code agentic workflows for real business use cases
• Identify and prioritise high-impact Agentic AI opportunities across business functions
• Plan, run and scale Agentic AI pilots into operational workflows
• Apply governance, risk and accountability frameworks for AI agents
• Lead the adoption of Human + Agent teams within the organisation
• Demonstrate leadership as an Agentic AI decision-maker

Capstone Phases:

1. Phase 1: Discover & Frame

2. Phase 2: Redesign with Agentic AI 

3. Phase 3: Prototype (No-Code)

4. Phase 4: Responsible AI 

5. Phase 5: Scale & Pitch

Introduction to Phase 3: Prototype (No-Code)

This phase focuses on building and testing a working prototype. You will implement one agent from your redesigned workflow and observe how it behaves under both normal and edge-case conditions.

The objective is not to build a production-ready solution, but to gain practical experience in building, testing and evaluating an Agentic AI workflow.

Scope guardrail 

You are building one agent end-to-end. Not three. Not the whole workflow. Just one: the Listing Generation Agent. It takes a product brief and produces an Amazon AE listing for a candle SKU. The other two agents from your Phase 2 blueprint stay on the roadmap. They are not in scope for the prototype.

Resources for this phase
Click here to download the resource for phase 3: Phase 3 Download  Phase 3

Note: n8n is provided as the reference implementation platform. You may use alternative workflow automation orchestration or agent-building tools, provided the solution delivers the same functionality and outputs. You may use Make, Zapier, Lindy, or any other no-code platform if you have prior experience with it. The starter template will not transfer, so you will need to build the equivalent structure yourself. If you do this, plan an extra 2-3 hours.

Your Task:

Step 1 – Workflow Overview
It should contain the following:

a) workflow architecture

(b) workflow execution 

(c) generated output. 

Step 2 – Test the Golden Case

Capture screenshots showing:

• Input
• Output

Step 3 – Test the Edge Case

Capture the screenshots showing:

• Input
• Prompt or agent configuration ​
• Generated output ​
  
  Document the behaviour

Self-Check Before Moving On

• Does the agent work end-to-end on the golden test case (file 10) without manual intervention?
• Did you capture how the agent behaves on the edge case (file 11)? You will need this in Phase 4.
• Is your prompt explicit enough that someone else could understand what it is trying to do?

When to Complete

Complete this phase alongside Modules 7–11.

Word Limit: 2000 

Estimated Duration: 6 Hours

Submission Note

• Use the same template to add responses that you downloaded in Phase 1. 
• Add your responses to phase 3 slides #12 to #19 of the capstone project template.
• You can add more slides to a task if required.
• You do not need to submit Phase 3 as a standalone assignment.
• Submit the completed capstone deck in Week 18.

This is a required assignment and counts towards programme completion.

Here is the rubric that will be used to grade your Phase 3 submission when you submit your completed capstone deck in Week 18.

Maximum points: 25