🤖 AI for Software Testers

Module 1: Prompt Engineering — "Where to Start!"

What is Prompting?

Prompting is the practice of writing clear, structured instructions to an AI model to get useful, accurate output. A good prompt sets context, defines the task, and specifies the desired format. For testers, this means the difference between vague, unusable AI output and production-ready test cases, bug reports, or automation scripts.

Steps to Learn Prompting

A practical progression for testers:

1. Start simple: Ask direct questions ("Write 5 test cases for a login form").
2. Add context & constraints: Include requirements, user stories, and acceptance criteria in your prompt.
3. Iterate on outputs: Refine the AI's response by giving feedback ("Make the steps more detailed", "Add negative test cases").
4. Use structured frameworks: Apply role-based and step-by-step prompts for complex tasks.

Zero-Shot vs Few-Shot Prompting

Types of Prompts

Prompting Frameworks: SWOT, STAR, CLEAR, PAR

Structured templates borrowed from business and writing to organize prompts effectively:

• SWOT (Strengths / Weaknesses / Opportunities / Threats) — for analysis tasks like risk assessment.
• STAR (Situation / Task / Action / Result) — for scenario-based answers and retrospective reporting.
• CLEAR (Concise / Logical / Explicit / Adaptive / Reflective) — a general prompt quality checklist.
• PAR (Problem / Action / Result) — similar to STAR, ideal for structured bug descriptions and incident reports.

How NOT to Use ChatGPT / AI Tools

Advanced Prompt Engineering for QA

Deeper techniques tailored to testing work:

• Prompt chaining: Multi-step test design — first generate scenarios, then expand each into detailed test cases, then generate automation code.
• System prompts: Enforce QA standards by setting persistent instructions like "Always include preconditions, test data, and expected results."
• Context injection: Combine prompts with real project context (requirements docs, existing test suites, API specs) for dramatically higher-quality output.

LLMs & Local Tools

What is an LLM?

A Large Language Model (LLM) is an AI system trained on massive text datasets to understand and generate human-like language, powering tools like ChatGPT, Claude, and Gemini. LLMs are the engine behind modern AI-assisted testing.

Try Different LLMs

Hands-on comparison of multiple models to understand their differing strengths for QA tasks:

Hosting DeepSeek / Open LLMs Locally

Running open-source LLMs on your own machine instead of cloud APIs — useful for data privacy, offline access, and avoiding API costs. Common tools include Ollama and LM Studio.

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull and run DeepSeek Coder model
ollama pull deepseek-coder:6.7b
ollama run deepseek-coder:6.7b

# Example: Ask it to write a Selenium test
# "Write a Selenium test for login validation"

Cursor AI & GitHub Copilot

AI-powered code editors and assistants that autocomplete, explain, and refactor test automation code in real time. Cursor is an AI-native IDE, while GitHub Copilot integrates into VS Code and IntelliJ as an inline suggestion engine.

Hallucination Detection & Factuality Testing

Techniques to identify when an AI model generates false or fabricated information ("hallucinations"):

• Cross-reference generated test steps against actual application behavior.
• Verify API endpoint names, HTTP methods, and status codes against the real API specification.
• Use ground-truth comparison: run AI-generated assertions against known expected values.
• Flag outputs that sound confident but cite non-existent methods, classes, or frameworks.

Module 2: Generative AI — "What Can It Build Me?"

QA & Automation Testing Principles (AI-Assisted Lens)

Foundational testing concepts — test levels, test types, the testing pyramid — revisited through an AI-assisted lens. AI accelerates but does not replace core QA fundamentals. It's a force multiplier for experienced testers.

Requirement Analysis with AI

Using AI to read requirement documents or user stories and extract:

• Testable conditions — functional and non-functional acceptance criteria.
• Edge cases — boundary values, error conditions, race conditions.
• Ambiguities — vague or contradictory requirements a human reviewer might miss.

Test Planning & Strategy with AI

Prompt AI to help draft a test plan (scope, objectives, resources, schedule) and a test strategy (overall approach: risk-based, automation-first, etc.). Use AI-generated templates as starting points:

Test Case Generation — Manual & Automated

Prompting AI to write manual test case steps from requirements, as well as generating automated test scripts (e.g., Selenium/Java) directly from those same requirements.

You are a senior QA engineer. Given the following user story:

"As a user, I want to reset my password via email so that
I can regain access to my account."

Generate:
1. 8 manual test cases (positive + negative) in a table
   with columns: TC_ID, Description, Steps, Expected Result
2. A Selenium Java test for the happy path using POM pattern

Bug Identification & Reporting with AI

Using AI to help spot inconsistencies from logs or screenshots and draft clear, well-structured bug reports with: steps to reproduce, expected vs actual behavior, severity classification, and environment details.

API Test Case & Script Generation

Using AI to generate API test cases and automation code across popular tools:

Test Data Management for APIs

Strategies for creating, managing, and cleaning up test data needed for API testing — including using AI to generate realistic but synthetic datasets (names, emails, transactions, addresses) that look production-like without using real PII.

Automated Reporting — Allure Report & Jenkins Integration

Setting up Allure Report for rich, visual test reports and integrating automated test runs into Jenkins CI/CD pipelines for continuous reporting with pass/fail dashboards, trend graphs, and failure categorization.

Performance & Load Testing with AI Prompts

Using AI prompts to help design performance/load test scenarios, generate test scripts (e.g., for JMeter or k6), and interpret performance results — identifying bottlenecks, response time thresholds, and concurrency limits.

Module 3: AI Agents & MCP Server — "What Can It Build?"

n8n Basics + AI Agents

n8n is an open-source workflow automation platform. You can embed AI "agents" (LLM-powered decision-making nodes) into automated workflows — for example, an agent that automatically triages incoming bug reports, categorizes them by severity, and assigns them to the right team.

Code Explanation & Review AI

Using AI to read existing automation/test code and:

• Explain what each section does in plain English.
• Review for code smells, anti-patterns, and best-practice violations.
• Suggest improvements — better locator strategies, wait handling, assertion clarity.

AI Code Error Identification

Having AI scan code or error logs/stack traces to pinpoint the root cause of failures faster than manual debugging. Feed it the exception, the test code, and the application context — it can often identify the exact issue and suggest a fix.

AI Code Optimization

Using AI to refactor and improve existing code for readability, performance, or maintainability without changing its functionality. Examples: extracting helper methods, applying design patterns, removing duplication, improving naming.

Generating & Updating Code (Java, REST Assured, etc.)

Prompting AI to write new automation code or update existing scripts in Java, REST Assured, and similar frameworks based on natural-language instructions like "Add retry logic to the API tests" or "Convert this TestNG test to JUnit 5".

MCP Server (Model Context Protocol)

What is MCP Server?

MCP (Model Context Protocol) is a standard that lets AI models connect to external tools and data sources (databases, APIs, file systems, CI/CD pipelines) in a structured, secure way. Think of it as "USB for AI" — a universal interface between LLMs and the real world.

Use Cases of MCP Server for Testers

Allow LLM to Talk to MCP

The technical setup for connecting a language model client to an MCP server so it can call external tools as part of its responses — enabling truly autonomous testing agents that can analyze, decide, and act.

Selenium AI Learning Plan

AI Project Setup (Maven, IntelliJ, TestNG, Allure)

Step-by-step setup of a Java automation project with AI assistance throughout:

<dependencies>
  <!-- Selenium WebDriver -->
  <dependency>
    <groupId>org.seleniumhq.selenium</groupId>
    <artifactId>selenium-java</artifactId>
    <version>4.20.0</version>
  </dependency>

  <!-- TestNG -->
  <dependency>
    <groupId>org.testng</groupId>
    <artifactId>testng</artifactId>
    <version>7.10.0</version>
  </dependency>

  <!-- Allure TestNG -->
  <dependency>
    <groupId>io.qameta.allure</groupId>
    <artifactId>allure-testng</artifactId>
    <version>2.25.0</version>
  </dependency>
</dependencies>

Sample Project Structure

selenium-ai-project/
├── src/
│   ├── main/java/com/project/
│   │   ├── pages/          # Page Object classes
│   │   ├── utils/          # Utilities (DriverFactory, ConfigReader)
│   │   └── base/           # BaseTest with setup/teardown
│   └── test/java/com/project/
│       ├── tests/          # Test classes
│       └── dataproviders/  # TestNG DataProviders
├── testng.xml
├── allure-results/
└── pom.xml

Code Optimization Patterns

Common design patterns used in automation frameworks — and how AI can help implement them correctly:

• Singleton Pattern: Single shared WebDriver instance across tests (DriverFactory).
• Base Test Class: Shared setup/teardown logic inherited by all test classes.
• Page Object Model: Encapsulate page elements and interactions in dedicated classes.
• Builder Pattern: Construct complex test data or API requests cleanly and readably.

Module 4: Advanced AI Tools & Integration

Web & Mobile Functional Automation Using AI

Using AI to accelerate writing and maintaining functional automation scripts for both web and mobile applications — including generating Appium scripts, handling dynamic locators, and self-healing test maintenance.

Advanced Problem Solving — E2E Web Projects

Tackling complex, end-to-end automation challenges on real web projects, using AI as a problem-solving aid for tricky scenarios: dynamic elements, iframe handling, shadow DOM, complex multi-step user flows, and cross-browser quirks.

SQL Query & Project Automation

Using AI to generate and optimize SQL queries for backend/database validation as part of an automated testing project — verifying data integrity, checking DB state after API calls, and validating stored procedures.

AI-Driven Synthetic Test Data Generation

Generating realistic but artificial test data (names, emails, addresses, transactions, credit card numbers) using AI to support testing without using real or sensitive production data. Supports GDPR/data-privacy compliance in test environments.

ReportPortal: AI-Driven Test Analytics

ReportPortal is a real-time reporting and analytics tool for automated tests. Its AI features auto-analyze test failures, detect patterns across runs, classify defects (product bug vs automation issue vs system issue), and surface trends to help teams focus on what matters.

DeepEval: Comprehensive LLM Testing Framework

DeepEval is a framework specifically designed for testing and evaluating LLM-based applications. It checks:

• Accuracy: Are the AI outputs factually correct?
• Relevance: Do responses match the query intent?
• Hallucination rate: How often does the model fabricate information?
• Toxicity & bias: Does the output contain harmful or biased content?

Bonus: Fix Your Resume, Craft It & Apply

ATS-Friendly Resume Creation

Using AI to review your resume against Applicant Tracking System (ATS) standards, score it for keyword density and formatting, and suggest concrete edits to improve pass-through rates. Most companies use ATS filters — your resume needs to pass the robot before reaching a human.

Resume Tailoring for Job Descriptions

Prompting AI to customize your resume's wording and emphasis to match the specific keywords and requirements of a target job posting — without fabricating experience.

Mock Technical Interviews (Java, SQL, Testing)

Using AI as a practice interviewer to simulate technical interview questions on Java, SQL, Selenium, API testing, and QA concepts — with detailed feedback on your answers, follow-up questions, and scoring.

LinkedIn Profile Optimization

Using AI to rewrite your LinkedIn headline, summary, and experience sections to be more compelling and keyword-optimized for recruiter searches. Include your testing tools, certifications, and quantified achievements.

Cover Letters, Cold Emails & Self-Introduction Scripts

AI-assisted drafting of cover letters, cold outreach emails to recruiters/hiring managers, and scripts for video introductions or elevator pitches — personalized to each target company and role.