Individual Assignment: Build and Evaluate an AI Chatbot
Marks: 10
Assignment Title
Designing a No-Code AI Chatbot for a Real Business Problem
Objective
This assignment evaluates your ability to design, build, test, and critically assess an AI-
enabled chatbot for a real brand or organization. You must demonstrate that the chatbot
addresses a meaningful customer problem, uses appropriate information, provides a coherent
conversation experience, and includes suitable safeguards and human handoff.
Task
Choose one real company, brand, public organization, or service provider and build a no-code
chatbot using a pljatform such as Voiceflow, Botpress, Custom GPT, Microsoft Copilot
Studio, any other no-code chatbot platform.
The chatbot should address one clearly defined use case.
Possible use cases include any of the following:
Product recommendation
Customer service
Lead generation
Booking assistance
Product discovery
Complaint triage
FAQ support
Onboarding
After-sales assistance
Educational guidance
Avoid creating a general chatbot that attempts to answer everything.
Assignment Requirements
1. Business Problem and Chatbot Objective
Briefly explain:
The chosen organization
The customer problem
The target users
The role of the chatbot
The business and customer outcome expected
Your chatbot must have a clear objective such as:
Improving resolution
Reducing customer effort
Increasing conversion
Supporting product discovery
Generating qualified leads
Improving customer satisfaction
2. Conversation and Experience Design
Design the chatbot’s conversation architecture.
Include:
Welcome message
Main user intents
Key questions asked by the chatbot
Response paths
Clarification or grounding mechanism
Closure message
Human handoff condition
Submit a simple flowchart or conversation map.
The chatbot should demonstrate:
Turn-taking
Turn initiation where appropriate
Context retention
Clear navigation
Recovery from unclear queries
3. Knowledge Base and Content Quality
Create or upload an appropriate knowledge base for the chatbot.
This may include:
Product information
FAQs
Service policies
Pricing
Booking rules
Return or cancellation policies
Contact and escalation details
Explain:
What information the chatbot uses
What information it should not invent
How the knowledge base supports accurate responses
How outdated information will be managed
4. Functional Chatbot Prototype
Build a functional prototype.
The chatbot should be able to complete at least three distinct user tasks.
Examples:
Recommend a suitable product
Answer a policy question
Collect a lead
Provide booking information
Escalate a complaint
Guide a user to the next step
The prototype should not consist only of static buttons. It should demonstrate some
meaningful conversational capability.
Submit:
Working chatbot link, where possible
Screenshots
Brief explanation of the build structure
5. Testing and Evaluation
Test the chatbot using at least five user scenarios.
Your test cases must include:
1. A normal query
2. An ambiguous query
3. An incorrect or unsupported query
4. A frustrated or dissatisfied user
5. A case requiring human handoff
For each test, report:
User input
Chatbot response
Whether the response was satisfactory
Error or limitation identified
Improvement made
6. Responsible AI and Managerial Reflection
Critically evaluate the chatbot.
Discuss:
Possibility of hallucination
Privacy and data collection
Bias or exclusion
Transparency that the user is interacting with AI
Situations where human intervention is necessary
One managerial risk
One improvement for future deployment
Conclude by answering:
Should this chatbot be deployed, piloted further, or rejected in its current form? Justify your
decision.
Submission Format
Each student must submit:
1. Assignment report: 1,200–1,500 words
2. Conversation-flow diagram
3. Working chatbot link, where available
4. Screenshots or a 3–5 minute screen recording
5. Test-case table
6. Reflection and deployment recommendation
Suggested file name:
Assignment1_AIM26-27_RollNumber
Important Instructions
The assignment is about solving a customer or organizational problem, not merely
demonstrating a chatbot tool.
Do not submit a chatbot based only on default templates.
Avoid vague prompts and generic responses.
Do not upload confidential or personal information into public chatbot platforms.
Unsupported claims must not be generated.
The chatbot must clearly state its limitations and provide a human handoff where
necessary.
Students will be evaluated on managerial relevance, conversation quality, testing
rigour, and responsible AI design, not on visual attractiveness alone.