Artificial Intelligence is rapidly transforming the way quality assurance is delivered. At Inspired Testing, we see AI not as a replacement for human judgment, but as a powerful accelerator of it.
This guide has been designed for testers who are using general AI tools — such as ChatGPT, Gemini, or Claude — rather than VeloAI. Its purpose is to help you harness AI responsibly, effectively and in alignment with Inspired Testing’s values of technical excellence, integrity and quality.
Through structured prompts and practical examples, this guide will show you how to:
- Analyse requirements more thoroughly
- Generate comprehensive and risk-based test cases
- Produce cleaner automation code and data sets faster
- Communicate results and insights more clearly
- Maintain full compliance with ISO 27001 and ethical AI standards
Each module has been developed to mirror the way Inspired Testing consultants work — combining deep testing expertise with practical efficiency. Every example prompt is designed to make AI collaboration part of your daily testing workflow, whether you’re working on web, mobile, performance, or enterprise systems.
Remember: AI is only as valuable as the precision and context of your prompts. The more specific, structured and ethical your inputs are, the better the outputs will be. Use this guide as both a training resource and a live reference in your projects.
Our goal: to enable every Inspired Testing professional to work smarter, faster and more effectively — with AI as a trusted partner in quality.
Module 1 – Getting Started & Prompt Engineering for Testers
Purpose
This module introduces testers at Inspired Testing to the principles of effective AI-assisted prompting. It is designed for those using general AI tools — such as ChatGPT, Claude, or Gemini — instead of VeloAI. It explains how to design, refine and structure prompts that help testers analyse requirements, generate test ideas and document findings while upholding the company’s ISO 27001-aligned data-handling standards.
1.1 The Role of AI in Testing
AI does not replace professional judgement.
It enhances productivity by:
- Generating structured test ideas in seconds
- Supporting rapid analysis of ambiguous requirements
- Assisting with automation or defect documentation
- Producing summaries and stakeholder-ready explanations
AI accelerates your thinking — it doesn’t replace it.
1.2 Responsible Use at Inspired Testing
All testers must comply with Inspired Testing’s ISO 27001 policy:
- Never paste client data, code, or credentials into public AI tools.
- Use synthetic or anonymised examples.
- Treat every AI output as a draft to be verified.
- Maintain human review and accountability for deliverables.
1.3 Anatomy of a Strong Prompt
| Element | Description | Example |
|---|---|---|
| Role | Define who the AI acts as | “You are a senior QA consultant specialising in API testing.” |
| Context | Supply feature or story details | “You are testing a login API with token-based authentication.” |
| Task | State what is needed | “List positive, negative and boundary test cases.” |
| Constraints | Impose limits or structure | “Output as a Markdown table (ID, Input, Expected Result).” |
1.4 Weak vs Strong Prompts
| Weak Prompt | Strong Prompt |
|---|---|
| “Write test cases for login.” | “You are a QA analyst validating a banking app’s login API. Generate 10 test cases covering positive, negative and boundary conditions. Include inputs, expected outputs and validation notes in a table.” |
| “Explain this bug.” | “You are a QA consultant preparing a defect summary. Rewrite this description for a client stakeholder; include business impact and probable root cause.” |
1.5 Prompting Techniques
A. Role-Based Prompting
You are a test lead at Inspired Testing reviewing user stories.
Identify five ambiguities and suggest clarifying questions for the business analyst.
B. Context-Rich Prompting
You are validating the requirement:
‘System must support password resets via email.’
List functional, negative and edge test scenarios.
C. Structured Output Prompting
Provide output as a Markdown table with columns:
Scenario ID | Input | Expected Result | Type (Positive/Negative)
D. Iterative Refinement
1. Start broad → “List test cases for a login page.”
2. Narrow → “Now focus on session-handling negatives.”
3. Format → “Convert the final set to Gherkin scenarios.”
1.6 Advanced Prompt Patterns
Chain of Thought
You are a QA specialist preparing for exploratory testing.
Step 1 – Identify primary user actions.
Step 2 – Derive test heuristics.
Step 3 – List edge cases grouped by risk level.
Constraint-Driven
Generate exactly 8 test scenarios for an online checkout:
4 positive + 4 negative, each under 25 words.
Few-Shot
Provide an example → the AI learns your format:
Example:
| ID | Scenario | Expected |
|----|-----------|-----------|
| TC01 | Valid email & password | Redirect to dashboard |
Now create 10 more in the same format.
1.7 Refining AI Output
Use follow-up prompts to evolve results:
- “Expand this to include boundary cases.”
- “Focus on security testing aspects.”
- “Summarise for management audience.”
- “Add a column for test-data requirements.”
1.8 Inspired Testing QA Prompt Framework
ROLE: [QA lead | Test analyst | Automation engineer]
CONTEXT: [Feature, API, requirement]
TASK: [Generate test cases | Write defect | Summarise report]
FORMAT: [Table | Gherkin | JSON | Summary]
CONSTRAINTS: [# of items, tone, length, focus]
Example
ROLE: Senior QA consultant at Inspired Testing
CONTEXT: Password-reset workflow for a banking app
TASK: Create 12 functional and negative test scenarios
FORMAT: Markdown table (ID, Scenario, Expected Result, Notes)
CONSTRAINTS: One scenario per step; note dependencies
1.9 Common Prompt Structures
| Goal | Prompt |
|---|---|
| Clarify requirements | “List ambiguous phrases and suggest clarifying questions.” |
| Generate test ideas | “List 15 exploratory ideas for a file-upload feature grouped by heuristic.” |
| Help with automation | “Convert this manual test into Playwright code with assertions.” |
| Defect writing | “Rewrite this issue clearly for the client’s defect tracker.” |
| Report summary | “Summarise this QA report for executives focusing on risk and trends.” |
1.10 Do’s and Don’ts
| Do | Don't |
|---|---|
| Use AI for brainstorming and drafting. | Paste client data or confidential artefacts. |
| Verify every output yourself. | Assume AI answers are factually accurate. |
| Use structured prompts for consistency. | Request opinions on clients or people. |
| Combine AI and human analysis. | Depend solely on AI judgement. |
1.11 Example Workflow
Prompt 1: List functional and boundary test cases for registration form.
Prompt 2: Add negative cases for missing mandatory fields.
Prompt 3: Convert to Gherkin format.
Prompt 4: Highlight top 3 high-risk scenarios.
1.12 Key Takeaways
- Clear intent + structure + iteration = useful results.
- Protect client data and confidentiality.
- AI enhances your testing intelligence — it doesn’t replace it.
- Review and refine everything you generate.
Module 2 – Requirements Analysis & Test Planning
Purpose
This module helps testers use AI tools to interpret requirements, expose ambiguities and create test-planning artefacts early in the lifecycle.
At Inspired Testing, QA consultants engage early with business analysts and developers — AI can act as a “junior analyst” that assists, not replaces, the human reviewer.
2.1 Analysing Requirements
Goal
Identify unclear or incomplete parts of a requirement and propose questions or acceptance criteria.
Example Requirement
“When users add products to the basket, the system should automatically calculate discounts and display updated totals in real time.”
Prompt Examples
You are a senior QA consultant reviewing an e-commerce user story.
Story: "When users add products to the basket, the system should automatically calculate discounts and display updated totals in real time."
Identify potential ambiguities or missing acceptance criteria.
Output as 'Observation → Clarification Needed' pairs.
Break the requirement into atomic, testable statements.
Classify each as Functional, Non-Functional, or UI-related.
List five clarification questions a QA would raise about discount logic, triggers, or thresholds.
Expected Output
A structured list of ambiguous phrases, missing rules (e.g., discount types, rounding behaviour) and a set of clarifying questions to raise in refinement.
2.2 Identifying Risks & Dependencies
Example Context
“The system must support both web and mobile checkouts using a shared API.”
Prompt Examples
You are preparing a test strategy.
Requirement: "The system must support both web and mobile checkouts using a shared API."
List functional, integration and performance risks that may impact testing.
Output a 2-column table: Risk | Mitigation / Test Consideration.
Identify dependent components and services that require validation
(front-end frameworks, API versions, database schema, external payment gateway).
Expected Output
A concise risk matrix highlighting concurrency, data consistency, API versioning and environment-specific performance issues.
2.3 Generating Acceptance Criteria
Example Requirement
“A user can reset their password via email if they forget it.”
Prompt Examples
Act as a QA consultant helping to refine acceptance criteria.
Convert this requirement into Gherkin-style Given/When/Then scenarios.
Include success and failure paths.
List acceptance criteria for password-reset:
- Valid email flow
- Invalid or expired link
- Rate-limit or throttling case
Expected Output
Scenario: Successful reset
Given a registered user requests a reset
When they click a valid link within 15 minutes
Then the system prompts for a new password and confirms change
Scenario: Expired link
Given a link older than 15 minutes
When the user opens it
Then show “Link expired” and offer resend
2.4 Coverage & Traceability
AI can help build simple traceability matrices ensuring all acceptance criteria are test-covered.
Prompt Examples
Create a traceability matrix mapping requirements → criteria → test ideas.
Columns: Req ID | Acceptance Criteria | Test Ideas | Notes.
Perform a gap analysis: compare these criteria with the parent requirement and highlight missing conditions.
2.5 Estimating Scope & Effort
Estimate QA effort for the password-reset feature.
Classify effort as Low / Medium / High and explain reasoning in three points.
Group test cases into Manual / Automated / Exploratory.
Suggest which to automate first based on stability and ROI.
2.6 Scenario Generation
Example Requirement
“The loan calculator should return results instantly and support values from £1 000 to £50 000.”
Prompt Examples
Design functional, boundary and negative tests for the loan calculator.
Include valid ranges, input errors and performance aspects.
Return a table: ID | Scenario | Input | Expected Result | Type.
Extend the list to include security and usability considerations.
Expected Output
| ID | Scenario | Input | Expected Result | Type |
|---|---|---|---|---|
| TC01 | Valid loan amount | £10 000 | Result returns instantly | Functional |
| TC02 | Below minimum | £999 | Error “Min £1 000” | Negative |
| TC03 | Upper boundary | £50 000 | Result ≤ 2 s | Boundary |
| TC04 | Invalid chars | “abc” | Validation error | Negative |
2.7 Story Refinement & Sprint Preparation
Summarise each backlog story in one sentence and note key QA focus.
Output a 3-column table: Story | Summary | QA Focus.
Review this backlog item and list unclear criteria or technical dependencies to raise during refinement.
2.8 Drafting Mini Test Strategies
Create a short test strategy for an insurance-quote system.
Sections: Objectives, Scope (in/out), Key Risks, Environments.
Format as Markdown headers.
List main QA deliverables and entry/exit criteria from the given requirements.
2.9 Key Takeaways
- Engage AI early – before sprint start.
- Treat AI as a structured reviewer.
- Verify all acceptance criteria with stakeholders.
- Maintain visible traceability between requirements and test artefacts.
Module 3 – Test Design & Test Data Generation
Purpose
This module teaches testers how to use AI tools to transform analysed requirements into robust, risk-based test cases and realistic test data.
At Inspired Testing, structured test design is at the heart of quality assurance — AI helps generate breadth quickly while leaving depth and prioritisation to human expertise.
3.1 From Requirement to Test Case
Goal: Convert clarified requirements into complete, traceable test cases.
Prompt Examples
You are a QA analyst designing tests for a loan-application workflow.
Create positive, negative and boundary scenarios covering input validation, calculations and error messages.
Return as a Markdown table with: ID | Scenario | Pre-conditions | Steps | Expected Result.
List at least 10 edge cases often missed in financial-form validation and mark which are High, Medium, or Low risk.
Translate the following acceptance criteria into full test cases using Given/When/Then format.
3.2 Functional Test Design
Generate test cases for an online checkout process including:
- Cart updates
- Payment gateway failures
- Voucher / discount application
- Timeout / session expiry
Output as a concise table grouped by test type (Functional | Integration | Error Handling).
You are testing an insurance-quote API.
Produce functional and error scenarios for the endpoint /quote.
Include boundary values, invalid data and authentication checks.
3.3 Exploratory and Heuristic Testing
AI can assist in creating test charters or exploratory checklists.
Prompt Examples
You are planning a 60-minute exploratory session on a new “File Upload” feature.
Generate a list of test ideas grouped by heuristic (Data, Stress, Usability, Error Handling, Security).
Suggest 10 exploratory ideas focusing on accessibility, localisation and browser-compatibility risks.
3.4 Negative and Boundary Testing
Create boundary-value test cases for an age-input field that accepts 18–65 inclusive.
Include just-inside, just-outside and invalid values.
Generate equivalence-partition test sets for a password field with the following rules:
- 8–16 characters
- At least 1 uppercase, 1 number, 1 symbol
- No spaces
3.5 Data-Driven Testing Prompts
AI can automate data-set generation for both functional and non-functional testing.
Prompt Examples
Generate 25 rows of synthetic test data for an insurance-policy form:
Columns: PolicyID, CustomerName, Age, PremiumAmount, StartDate, EndDate.
Include edge cases like future start dates or negative premiums.
Produce JSON payloads for the “CreateUser” API including:
- Valid user
- Duplicate email
- Missing mandatory field
- SQL-injection attempt
Provide CSV-ready data for load testing a login endpoint (username | password | expected status).
3.6 Domain-Specific Examples
Banking / Finance
Design test scenarios for fund transfer between accounts covering limits, currency differences and authorisation failures.
Retail / E-Commerce
Generate 15 test cases for coupon redemption combining product categories, cart values and expiry rules.
Healthcare
Create data combinations for patient registration with required fields, optional insurance details and validation of medical-record IDs.
3.7 Performance and Resilience Data
Generate 100 unique payloads to simulate simultaneous loan-applications.
Each payload must vary income, loan amount and duration.
Return as JSON array suitable for JMeter.
Suggest 10 performance-testing scenarios focusing on concurrency and degradation points for a REST API.
3.8 Test Case Optimisation and Prioritisation
Given a list of 40 test cases, ask the AI to group by risk and recommend top 10 for smoke testing.
Classify tests into must-run vs nice-to-have based on customer impact and defect probability.
Output as table: ID | Scenario | Priority | Rationale.
3.9 Output Formatting and Traceability
Encourage structured outputs that can plug into management tools.
Format your test cases for export to Azure DevOps or TestRail.
Columns: Title | Preconditions | Steps | Expected Result | Priority.
Generate a Requirement-to-Test mapping document showing IDs and relationships.
3.10 Key Takeaways
- AI accelerates coverage creation but does not replace design judgement.
- Focus prompts on inputs → outputs → risk.
- Always validate generated data before use.
- Maintain traceability between requirements, test cases and test data.
Module 4 – Test Automation & AI-Assisted Scripting
Purpose
This module demonstrates how testers can use AI tools to accelerate automation, review scripts and optimise frameworks while maintaining security and code quality.
At Inspired Testing, automation is not a goal in itself — it is a multiplier of human insight. AI can help generate scaffolding and refactor code, but testers remain accountable for design, maintainability and robustness.
4.1 Generating Automation Code Safely
AI can rapidly create code snippets for common automation frameworks such as Cypress, Playwright, Selenium, REST Assured, or Postman.
However, testers must always review generated code for accuracy, data safety and compliance with client standards.
Prompt Examples
You are a QA automation engineer using Playwright with TypeScript.
Generate a test script for the login page that:
- Opens the login URL
- Enters credentials
- Verifies redirect to the dashboard
Include clear comments.
Write a Selenium-Java test for verifying error messages on invalid email input.
Add WebDriver setup and teardown.
Convert this manual test into a Cypress script.
Manual steps:
- Navigate to the cart
- Apply a voucher code
- Verify discount is applied
Expected Output
A complete, commented test skeleton ready for review and refactoring.
4.2 Improving and Refactoring Automation Code
Review the following Selenium test code and suggest:
- Simpler locators
- Better wait strategies
- Improved assertions
Optimise this Cypress test for readability and reduce duplication using page objects.
Explain how to parameterise this test to support multiple browsers.
Use AI as a code reviewer to improve style consistency, readability and maintainability.
4.3 Framework Design Assistance
AI can generate scaffolding for framework layers, reducing repetitive setup work.
Prompt Examples
Design a basic page-object model structure for a React web app using Playwright.
List recommended folder names and file organisation.
Suggest a naming convention for test classes and methods in a Java-TestNG framework.
Generate a template for logging and reporting integration using Allure.
4.4 API Automation
You are testing a bank transfer API.
Generate Postman or REST Assured requests for the following scenarios:
- Valid transfer
- Invalid account number
- Insufficient balance
- Unauthorised token
Produce a JSON schema validation snippet for the /transfer endpoint response.
Write a Python requests-based API test that verifies HTTP status codes and response fields.
4.5 Data-Driven and Parameterised Tests
Demonstrate how to parameterise Cypress tests using external CSV data.
Include one positive and one negative example.
Generate JUnit test data parameterisation for login combinations.
Fields: username, password, expectedStatus.
Create a data-driven test for a search API that reads queries from a CSV file and verifies response time < 2 s.
4.6 Continuous Integration and Pipeline Support
You are integrating automated tests into an Azure DevOps pipeline.
Generate YAML steps to:
- Install dependencies
- Run Cypress tests
- Publish Allure results
Create a GitHub Actions workflow for running Python API tests on every pull request.
Include test summary reporting step.
Summarise how to use AI to analyse failed pipeline logs and identify flaky tests.
4.7 Maintenance and Self-Healing Assistance
Review this failed test log and suggest three possible causes for the timeout.
Recommend a more resilient wait condition.
Suggest how to implement self-healing locators using AI tools or dynamic selectors.
Analyse this suite’s execution times and recommend tests to run in parallel to reduce build duration by 30 %.
4.8 Cross-Browser and Device Testing
Generate Playwright configuration for running tests across Chrome, Edge and Firefox.
Create mobile-emulation scenarios for Chrome DevTools covering iPhone 14 and Galaxy S22 viewports.
List five test cases that must run on both mobile and desktop platforms to validate responsiveness.
4.9 Security and Compliance Within Automation
Highlight five security considerations to check before committing AI-generated automation code to a client repository.
Explain how to mask test data or use secure secrets storage within automated pipelines.
4.10 Key Takeaways
- Use AI to accelerate automation, not to bypass engineering rigour.
- Always review generated code for accuracy, reusability and security.
- Leverage AI for refactoring, data-driven design and CI/CD optimisation.
- Keep human QA oversight central to Inspired Testing’s automation ethos.
Module 5 – Reporting, Defect Management & Communication
Purpose
This module shows how testers can use AI to improve communication, produce concise reports and enhance defect documentation.
At Inspired Testing, clarity and consistency in QA communication are essential — AI should help structure and refine, but never distort, the facts.
5.1 Generating Test Summaries
Prompt Examples
You are a QA consultant preparing a daily execution summary.
Summarise these test results:
- 125 tests executed
- 9 failed
- 4 blocked due to environment issues
Output: concise summary + top 3 risks.
Generate a test summary email for a project manager, highlighting:
- Tests executed and pending
- Major blockers
- Next steps for tomorrow
Tone: concise, professional and non-technical.
5.2 Executive Reporting
AI can adapt technical details into business-friendly summaries.
Prompt Examples
Summarise the following QA report for an executive audience.
Focus on risk, release readiness and trends rather than technical detail.
You are a QA lead preparing a weekly dashboard narrative.
Summarise pass/fail trends and defect closure rate in five bullet points.
Use AI to translate QA data into actionable, business-level insight.
5.3 Defect Documentation
Prompt Examples
You are a QA tester writing a defect report.
Draft a clear title, description and reproduction steps for:
“App crashes when saving profile photo larger than 5 MB.”
Include expected vs actual result and environment info.
Rephrase this defect summary in client-friendly language:
'NullPointerException at UserServiceImpl.java line 142'
Provide reproduction steps and add a short business impact summary for escalation.
5.4 Root-Cause & Trend Analysis
Prompt Examples
Analyse this defect list and group by root cause: Requirements, Code, Data, Environment, Other.
Summarise which category contributes most to total defects.
Given 10 recurring issues from the past 3 sprints, list 3 preventive measures for each.
5.5 Release Readiness Assessment
Prompt Examples
You are preparing a QA sign-off summary.
Generate 5 bullet points covering:
- Test completion %
- Outstanding defects
- Known risks
- Mitigation actions
- Go/no-go recommendation
Summarise release readiness for executives:
Provide 1-paragraph summary + traffic-light risk rating (Green/Amber/Red).
5.6 Communication with Stakeholders
Prompt Examples
Rewrite this QA update for a non-technical client stakeholder.
Keep tone neutral, focus on business impact and mitigation.
Draft a short Teams message updating the team on test progress and blockers.
Generate 3 ways to phrase sensitive feedback about recurring environment instability.
5.7 Visual Reporting Assistance
Prompt Examples
Convert this test execution table into a written summary.
Highlight top 3 failure trends.
Suggest 5 chart types that best visualise test progress and defect ageing for management.
5.8 Retrospective and Lessons Learned
Prompt Examples
Create a short “lessons learned” summary from the following QA project notes.
Group by Process, Tools, Communication and Quality.
Generate a retrospective summary slide text for QA contribution to sprint review.
5.9 Key Takeaways
- Use AI to enhance QA communication, not to embellish results.
- UKeep tone neutral, factual and business-oriented.
- UAlways validate AI-generated summaries against actual metrics.
- UClear communication builds trust with clients and stakeholders.
Module 6 – Advanced Testing Domains (Performance, Security, Accessibility)
Purpose
This module explores how testers can use AI to prepare for specialised testing areas — Performance, Security and Accessibility.
At Inspired Testing, these disciplines require domain expertise, tools and analytical depth. AI enhances preparation and documentation, helping testers think broadly, identify risk and create structured test ideas.
6.1 Performance Testing
Prompt Examples
Generate 10 performance-testing scenarios for an e-commerce checkout.
Group by Load, Stress, Spike and Endurance testing.
You are a QA performance specialist preparing a JMeter test plan.
List key metrics (Throughput, Response Time, Error %, CPU Utilisation).
Provide expected thresholds for a medium-traffic retail site.
Suggest data sets to simulate 100 concurrent users performing mixed operations (browse, add to cart, checkout).
Expected Output
A table grouping performance test ideas by type, including metrics and expected outcomes:
| Test Type | Scenario | Metric | Target |
|---|---|---|---|
| Load | 100 concurrent users | Response time | < 2 s |
| Stress | Gradual ramp-up to 1 000 users | Error rate | < 1 % |
| Spike | Sudden burst of 500 users | System recovery time | < 10 s |
| Endurance | 12-hour steady load | Memory usage growth | < 5 % |
6.2 Analysing Performance Results
Prompt Examples
Given response time metrics, summarise bottlenecks and likely causes.
Group by Server, Database, or Front-End.
Suggest 3 optimisation recommendations for endpoints exceeding SLA.
Expected Output
A structured summary categorising bottlenecks by layer and prioritising optimisations, e.g., “API payload compression” or “indexing of slow queries.”
AI helps testers move from raw data to actionable recommendations.
6.3 Security Testing
List common vulnerabilities to test in a login API according to OWASP Top 10.
Include test ideas for each vulnerability.
You are testing a financial application.
Generate 10 test ideas for verifying secure data handling, focusing on:
- Authentication
- Authorisation
- Session management
- Data encryption
Write a checklist for API security testing covering headers, tokens and payload validation.
Expected Output
| Vulnerability | Test Idea | Expected Outcome |
|---|---|---|
| Injection | Try ' OR 1=1-- in login field | Rejected safely |
| XSS | Submit <script>alert(1)</script> | Escaped output |
| Broken Authentication | Replay expired token | Request denied |
| Sensitive Data Exposure | Inspect network requests | Data encrypted |
6.4 Security Risk Review
Prompt Examples
Analyse this system architecture description and list possible security weaknesses.
Output as table: Component | Potential Vulnerability | Recommended Check.
Review this code snippet for potential injection vulnerabilities.
Provide reasoning and suggested fix.
Expected Output
| Component | Vulnerability | Recommended Check |
|---|---|---|
| Login API | Weak password policy | Enforce complexity and expiry |
| Database | SQL injection | Use parameterised queries |
| Session Token | Predictable pattern | Use UUIDs + short expiry |
6.5 Accessibility Testing
Prompt Examples
Generate a checklist based on WCAG 2.2 Level AA for web accessibility.
Include items for colour contrast, keyboard navigation and ARIA roles.
Suggest exploratory test ideas for accessibility validation using NVDA or VoiceOver.
You are reviewing a web form for compliance.
List 10 accessibility issues a tester should check manually.
Expected Output
| Category | Example Check | Tool / Method |
|---|---|---|
| Colour Contrast | Text contrast ratio ≥ 4.5:1 | Chrome DevTools / Axe |
| Keyboard Navigation | Tab order logical | Manual test |
| Screen Readers | Labels announced properly | NVDA / VoiceOver |
| Forms | Error messages accessible | Inspect DOM attributes |
6.6 Usability and Localisation
Prompt Examples
List usability heuristics (Nielsen’s 10) and generate 2 test ideas for each.
Generate test ideas for localisation testing of a multilingual app covering language fallback, date/time formats and encoding.
Expected Output
AI provides a matrix mapping usability heuristics to examples, such as “Visibility of system status → Display progress indicators,” and localisation cases like “Currency symbols adapt by region.”
6.7 Key Takeaways
- Use AI to plan and document specialised testing approaches.
- AI assists in checklists and thought organisation — it does not replace certified tools or security expertise.
- Maintain compliance with OWASP, WCAG and client security policies.
- For every AI-generated test idea, ensure a qualified tester validates its accuracy and feasibility.
Module 7 – Ethical Use, Confidentiality & ISO-Aligned AI Practice
Purpose
This module explains how testers at Inspired Testing must handle AI responsibly.
AI can enhance efficiency, but its misuse can expose sensitive data or misrepresent outputs.
By following ethical principles and aligning with ISO 27001 controls, testers ensure client trust and compliance.
7.1 Data Confidentiality
- Never paste client code, credentials, or production data into public AI tools.
- Use synthetic or masked data when demonstrating functionality.
- Verify whether the AI model stores prompts or responses; avoid tools that retain them.
- Prefer enterprise or private deployments for sensitive projects.
- Retain auditability: record prompt history in secured repositories when appropriate.
7.2 Ethical Prompting
Prompt Examples
List ethical considerations testers must follow when using AI for client work.
Group by: Data Privacy, Intellectual Property, Transparency.
Write a short policy statement describing acceptable use of AI in QA documentation.
Expected Output
A concise table mapping categories to practices:
| Category | Example Guideline |
|---|---|
| Data Privacy | No real user data or credentials in prompts |
| Intellectual Property | Respect client IP and licensing when referencing code |
| Transparency | Declare when AI contributed to deliverables |
7.3 Bias and Misuse Prevention
Prompt Examples
Explain how AI output bias could affect defect prioritisation or risk assessment.
Suggest safeguards.
You are a QA lead defining review steps for AI-generated content.
Create a checklist to verify accuracy, neutrality and confidentiality compliance.
Checklist Example
| Review Aspect | Check Action |
|---|---|
| Accuracy | Cross-verify facts against requirements |
| Neutrality | Remove subjective wording or bias |
| Confidentiality | Ensure data masking and anonymisation |
7.4 ISO 27001 Alignment
| ISO Domain | QA Context | Best Practice |
|---|---|---|
| Access Control | Limiting AI tool usage by role | MFA, logging, authorisation |
| Asset Management | Tracking prompt/response artefacts | Store securely in SharePoint |
| Data Protection | Handling of test data | Always anonymise or mask |
| Supplier Management | Using third-party AI vendors | Validate SLAs, data-handling terms |
Think of every prompt as data leaving your laptop — treat it with the same care as production information.
7.5 Communication and Transparency
- Always disclose when AI was used to draft, summarise, or analyse content.
- Maintain a human approval step before any AI-generated material reaches a client.
- Encourage open discussion of AI usage within teams — transparency drives accountability.
7.6 Continuous Ethics Review
Draft a quarterly checklist for reviewing AI usage across projects.
Include categories: Compliance, Data Handling, Training, Client Disclosure.
Create a one-paragraph ethics statement suitable for inclusion in a project QA plan.
7.7 Key Takeaways
- Protect client and company data above all else.
- Verify, review and anonymise before sharing.
- Align every AI action with ISO 27001 controls.
- Disclose AI assistance in deliverables.
- Uphold Inspired Testing’s reputation for trust, integrity and quality.
Module 8 – Prompt Library & Cheat Sheet (Quick Reference)
Purpose
This final module provides a quick-access prompt library for testers using general AI tools in their daily QA work.
It includes ready-to-use examples grouped by activity type — requirements, design, automation, reporting and compliance — for consistent, professional prompting across Inspired Testing teams.
8.1 Requirements Analysis
| Objective | Example Prompt |
|---|---|
| Clarify ambiguity | “Identify unclear or incomplete parts of this user story and list five clarifying questions.” |
| Derive acceptance criteria | “Convert this story into Given/When/Then acceptance tests.” |
| Identify risks | “List top five functional and integration risks for this feature.” |
| Analyse dependencies | “Identify external systems or APIs that could affect test planning.” |
8.2 Test Design
| Objective | Example Prompt |
|---|---|
| Generate boundary tests | “List boundary and equivalence-partition cases for this input field.” |
| Create exploratory ideas | “Suggest ten exploratory test ideas for this module grouped by heuristic.” |
| Negative test coverage | “Create five negative test cases focusing on input validation and data format.” |
| Prioritisation | “Group test cases by risk and business impact, marking each as High, Medium, or Low.” |
8.3 Test Data Generation
| Objective | Example Prompt |
|---|---|
| Generate synthetic data | “Create 20 rows of anonymised test data for an insurance policy form.” |
| Create JSON payloads | “Produce sample API requests for CreateUser endpoint including valid and invalid inputs.” |
| CSV for automation | “Generate CSV-ready data for login endpoint (username, password, expected result).” |
| Performance testing data | “Simulate 100 concurrent user sessions with varied input data for load testing.” |
8.4 Automation
| Objective | Example Prompt |
|---|---|
| Script generation | “Generate a Playwright test for login that asserts a successful redirect to dashboard.” |
| Refactoring | “Optimise this Selenium test for readability and maintainability using the Page Object Model.” |
| CI/CD assistance | “Write a GitHub Actions YAML snippet to run Cypress tests and publish reports.” |
| Code review | “Review this Cypress script and suggest improvements in waits, assertions and reusability.” |
8.5 Reporting and Communication
| Objective | Example Prompt |
|---|---|
| Daily summary | “Summarise today’s test execution (passed, failed, blocked) in five bullet points.” |
| Executive summary | “Summarise this QA report for executives focusing on risk, readiness and quality trends.” |
| Defect clarity | “Rewrite this defect for a client stakeholder, adding business impact and reproduction steps.” |
| Lessons learned | “Draft a QA retrospective grouped by Process, Tools and Communication.” |
8.6 Performance, Security, Accessibility
| Area | Example Prompt |
|---|---|
| Performance | “Generate load and stress test scenarios for checkout workflow with metrics and thresholds.” |
| Security | “List API test ideas for OWASP Top 10 vulnerabilities.” |
| Accessibility | “Write a checklist based on WCAG 2.2 AA covering colour contrast, keyboard use and ARIA labels.” |
8.7 Ethics and ISO Practice
| Area | Example Prompt |
|---|---|
| Confidentiality | “List five rules for handling client data securely when using AI tools.” |
| Review process | “Create a QA checklist for verifying accuracy and neutrality of AI-generated content.” |
| Transparency | “Write a statement acknowledging responsible AI assistance in deliverables.” |
| Audit readiness | “Generate a brief ISO 27001 compliance summary for AI-assisted QA workflows.” |
8.8 AI Enhancement & Self-Learning
| Area | Example Prompt |
|---|---|
| Continuous learning | “Suggest five ways AI can help testers upskill in exploratory and automation testing.” |
| Personal improvement | “Generate a weekly self-assessment checklist for measuring AI-assisted QA productivity.” |
| Knowledge sharing | “Write three LinkedIn-style insights about how AI improves QA efficiency.” |
8.9 Key Takeaways
- This prompt library is meant to accelerate daily testing work.
- Use it as a starting point — adapt each prompt to your specific project or domain.
- Always review AI output for technical and contextual accuracy.
- Maintain Inspired Testing’s high bar for professionalism, clarity and ethical standards.
Module 9 – Prompt Expansion Appendix
Deepening Context, Iteration and Precision
Purpose
This appendix provides advanced prompt-building patterns for testers who want to go beyond the quick examples in the Prompt Library.
It demonstrates how to create layered, context-rich prompts that follow Inspired Testing’s standards of completeness, traceability and professional communication.
9.1 The Structured Prompt Framework
Use this five-part template to build reliable, repeatable prompts.
| Element | Description | Example |
|---|---|---|
| Role | Define the AI’s testing perspective | “You are a senior QA consultant specialising in web accessibility.” |
| Context | Describe the system, feature, or data | “Testing the shopping-cart API for concurrency and pricing accuracy.” |
| Task | Specify the exact activity | “Generate 12 test cases including boundary and negative scenarios.” |
| Format | Tell the AI how to structure results | “Output as Markdown table: ID | Scenario | Input | Expected Result | Notes.” |
| Constraints | Impose limits or quality targets | “Keep descriptions under 20 words; highlight security-critical cases.” |
Full Example
ROLE: QA consultant for financial-loan portal
CONTEXT: Validating interest-rate calculator and error messages
TASK: Generate 15 functional and boundary test cases
FORMAT: Markdown table with 5 columns (ID | Scenario | Input | Expected | Risk)
CONSTRAINTS: One scenario per row; prioritise regulatory compliance conditions
9.2 Layered Prompting (Progressive Refinement)
Encourage testers to iterate rather than expecting perfect results in one go.
| Step | Goal | Example Prompt |
|---|---|---|
| 1 – Draft | Generate the baseline | “List functional and negative test cases for password reset.” |
| 2 – Refine | Add missing coverage | “Expand to include boundary and error-handling scenarios.” |
| 3 – Structure | Standardise the format | “Convert the list to a table with columns ID | Scenario | Input | Expected Result.” |
| 4 – Validate | Ask AI to self-check | “Review the table for duplication and missing edge cases.” |
| 5 – Summarise | Produce stakeholder view | “Summarise key risks and next steps in plain language.” |
Tip: Use the same chat thread to preserve context and let the AI build on its previous output logically.
9.3 Context Expansion Prompts
AI quality scales with the amount of context you provide.
Use these starter prompts to add richness:
Before generating tests, ask me any clarifying questions about:
- System purpose
- Input ranges
- Business rules
- Data dependencies
Here are the system details: [paste summary]
Using that, list assumptions you need clarified before generating test cases.
Create a list of environmental or integration factors that could affect test design.
9.4 Output-Control Techniques
Tell the AI exactly what format you expect.
Tabular Outputs
Provide results as a Markdown table with columns:
ID | Scenario | Input | Expected Result | Type (Positive/Negative)
Narrative Outputs
Write a 3-paragraph executive summary of the above results:
- Paragraph 1: Scope and objective
- Paragraph 2: Key findings
- Paragraph 3: Risks and recommendations
JSON / Code Outputs
Return the test cases as a JSON array with keys:
id, title, input, expected, priority
9.5 Meta-Prompts (Prompts About Prompts)
Teach testers to improve their own prompting quality.
Review my prompt and rate its clarity on a scale of 1–10.
Suggest two improvements for context and one for format.
Rewrite this prompt using the Inspired Testing structured format (Role, Context, Task, Format, Constraints).
Evaluate whether this prompt risks exposing client data or proprietary information.
If yes, rewrite safely using anonymised placeholders.
9.6 Combining Prompts Across Modules
Example compound workflow:
- Start with Module 2 (Requirements) → Generate acceptance criteria.
- Feed output into Module 3 (Test Design) → Expand into functional and negative cases.
- Use Module 4 → Generate automation code skeleton.
- Apply Module 5 → Create defect or summary narrative.
Prompt Example
Use the acceptance criteria below to generate functional and negative test cases.
Then convert them into Playwright scripts with comments.
Finally, produce a 5-bullet QA summary for management.
This teaches testers to chain AI outputs safely and efficiently.
9.7 Prompt Quality Checklist
| Category | Self-Check Question |
|---|---|
| Context | Have I explained what system or feature is under test? |
| Intent | Does the AI know what outcome I want (test cases, script, summary)? |
| Structure | Did I specify how I want results formatted? |
| Constraints | Did I limit quantity, tone, or detail appropriately? |
| Ethics | Did I remove any client or personal data before prompting? |
| Validation | Did I ask for AI to review or summarise its own output? |
9.8 Example: Complete End-to-End Prompt
ROLE: QA automation lead for an airline booking API
CONTEXT: Testing seat-selection and payment endpoints (REST, JSON)
TASK: Generate 20 functional, boundary and error test cases
FORMAT: Markdown table with ID, Endpoint, Input, Expected Result, Type
CONSTRAINTS: Include at least 3 authentication errors and 2 performance-related tests
Follow-Up Refinement
Now rewrite those tests as Gherkin scenarios with clear Given/When/Then syntax.
Then create a 5-point summary of high-risk areas.
9.9 Key Takeaways
- Structure first, detail second. A good prompt starts with context and ends with formatting.
- Iterate logically. Each refinement adds precision.
- Review ethically. Every prompt must respect data confidentiality.
- Adopt as habit. Consistent prompting discipline equals consistent quality.
Inspired Testing Standard: Treat AI like a junior tester — guide it clearly, review its work and help it improve.
