Working through this module?0 of 19 done
Module 02 · Page 3 of 20 · 7 sections

Requirements and Use Case Analysis

How to turn a vague idea into functional, non-functional and business requirements, worked through for an AI chatbot.

01Introduction to AI System Design07RAG System Design17End-to-End Project Design
2.1

Why Requirements are Important?

  • Help to understand what the system needs to achieve.
  • Define functional and non-functional requirements.
  • Identify constraints, performance goals and stakeholders.
  • Ensure the system is scalable, reliable and cost-effective.
2.2

Types of Requirements

1
Functional Requirements
What the system should do, e.g. process user queries, return accurate responses, support multiple data sources.
2
Non-Functional Requirements
How the system should perform, e.g. low latency, high availability, scalability, security, cost efficiency.
3
Business Requirements
Business goals and success metrics, e.g. improve customer support, reduce manual work, personalize recommendations.
2.3

Example: AI Chatbot System

Users
User QueriesText, voice, document, image
AI Chatbot System
ResponseAccurate, relevant, cited source, real-time
Internal DocsPDF, docs
DatabaseFAQs, policies
Web SearchReal-time
The chatbot answers from several knowledge sources rather than from the model alone.
2.4

Functional Requirements (Use Cases)

  1. 1User can ask natural language questions.
  2. 2System retrieves relevant information from internal documents.
  3. 3System generates accurate and contextual responses using an LLM.
  4. 4Supports multiple input types (text, file, image).
  5. 5Response should include source references.
  6. 6Support for chat history and follow-up questions.
2.5

Non-Functional Requirements

  1. 1Low latency (e.g. under 2 seconds).
  2. 2High availability (e.g. 99.9% uptime).
  3. 3Scalable to handle a large number of users.
  4. 4Secure access (authentication and authorization).
  5. 5Handle large documents and multiple data sources.
  6. 6Cost optimization (use caching, efficient models).
  7. 7Monitoring, logging and error handling.
Related11Reliability, Fault Tolerance and High Availability12Security, Privacy and Responsible AI13Monitoring, Logging and Observability
2.6

Stakeholders

1
End Users
Employees / customers.
2
Business Team
Define goals and success metrics.
3
Data Team
Provide and maintain data sources.
4
Engineering Team
Build and deploy the system.
5
Operations Team
Monitor and maintain.
6
Security Team
Ensure data privacy and compliance.
2.7

Success Metrics

1
Response Accuracy
Relevant and correct answers.
2
Latency
Average response time.
3
User Satisfaction
Feedback rating.
4
Adoption Rate
Active users.
5
Cost per Query
What each answer costs to produce.
6
System Reliability
Uptime and error rate.
Related14Evaluation and Testing