Creating Automated Tests with AI
Cory House
@housecor | reactjsconsulting.com
Demos: Implementing tests via AI
Unit testing
Vitest + Copilot
Bun + Copilot
E2E Testing
Playwright + Copilot
Playwright + ChatGPT requirements + Copilot
Playwright + ChatGPT requirements + Copilot + Inline instructions
Playwright + ChatGPT requirements + Copilot + Central instructions
Playwright + ChatGPT requirements + Copilot + Central instructions + Audit + Stryker
Applitools Autonomous
Visual Testing
Applitools Autonomous
Use requirements to generate tests.
No Requirements doc?
Generate via LLM!
Use requirements to generate tests.
Code Coverage
Measuring how much we’re testing
Code coverage sounds useful.
It’s a trap.
15
I’ll test 100% of the code.
@housecor
Make the tool say 100%.
When a measure
becomes a target,
it ceases to be a
good measure.
Goodhart’s Law
Autonomous self-maintaining website testing
✓ Easy one-click setup
Everything is included out of the box,
from devices & browsers to test scheduling.
✓ Automatic website discovery
Automatically crawls and finds all the pages
in your website given its URL.
✓ Comprehensive coverage
Automatically reports of any HTTP issue,
content change or rendering difference
✓ Adjusts to website changes
Automatically detects and reports new,
missing or faulty pages on every test run.
Interactive test authoring in plain English
✓ Author tests in plain English
No coding or locators skills required. Easiest
to understand and maintain.
Interactively record test steps
Speed up authoring by recording steps in
English, and easily refine manually if needed.
✓ Debug tests interactively
From setting breakpoints to step by step
execution. Do everything a code IDE can do.
✓ 100% SaaS
No need to install a local recorder. Record
and playback tests exactly as they will run.
Functional data driven testing
Robust textual assertions
Verify everything you can do with coded
tests from string formats to numeric values.
✓ Test parameterization
Execute tests with explicit arguments or per
row in a test arguments dataset.
✓ Capture data in variables
Extract value fragments and use them in
subsequent test steps.
✓ Execute JS code
When you need to describe complex logic,
reuse a utility or manipulate the page.
Visual assertions powered by VisualAI
✓ Complete coverage
Complete visual and functional coverage
across multiple devices and browsers.
✓ Fewer test steps
80% less test steps to author and maintain
for the entire lifetime of the test.
✓ Scalable maintenance
Accept or reject differences once across all
pages, devices and browsers
Cutting edge computer vision
Accurately mimics the human eye. 10+ dev
years / 4B+ tests. Dynamic pages.
API testing
Initiate HTTP requests
Custom authentication, reset app state, data
cleanup, or verify the backend state.
✓ Fully configurable
Specify HTTP verbs, cookies, headers, and
body. Use variables and parameters.
✓ Capture response details
Verify response details or store it in variables
to use in subsequent test steps
✓ Capture application requests
Capture outbound client requests and
responses to verify or use their data.
LLM assisted test authoring
Automatic error correction
Automatically corrects syntax and grammar
mistakes
✓ Automatic step simplification
Automatically splits complex steps to multiple
simpler ones
✓ Automatic step disambiguation
Automatically disambiguates step to clarify
how the step will be executed
✓ Intent based testing preview
Automatically generate test steps and test
data given a test goal.
LLM intent based testing
✓ Automatic error correction
Automatically corrects syntax and grammar
mistakes
✓ Automatic step simplification
Automatically splits complex steps to multiple
simpler ones
✓ Automatic step disambiguation
Automatically disambiguates step to clarify
how the step will be executed
Intent based testing preview
Automatically generate test steps and test
data given a test goal.
Summary
Playwright + ChatGPT + Copilot + Instructions
• Use requirements or generate a requirements doc via LLM.
• Tell the LLM to use the requirements doc to generate tests.
• Audit your markup to assure it’s accessible.
• Iterate. Tell the LLM about your standards (i.e. use Playwright’s recommended locators).
• Work through a feature at a time to keep scope reasonable.
• Use LLM instructions file (to streamline prompts and standardize output.
Applitools Autonomous
• Provide URL and it automatically crawls and generates.
• Capture data in variables for reference later.
• Call APIs and capture responses to validate expected server-side data.
• Can generate, host, and administer reliable visual tests too.
Want to explore AI-powered test
creation with Applitools Autonomous?
Schedule a call with a
Testing Specialist
applitools.info/v2j
Cory House | @housecor
Consulting: reactjsconsulting.com
Courses: pluralsight.com and dometrain.com
Questions?
Cory House | @housecor
Consulting: reactjsconsulting.com
Courses: pluralsight.com and dometrain.com

Creating Automated Tests with AI - Cory House - Applitools.pdf

  • 1.
    Creating Automated Testswith AI Cory House @housecor | reactjsconsulting.com
  • 2.
    Demos: Implementing testsvia AI Unit testing Vitest + Copilot Bun + Copilot E2E Testing Playwright + Copilot Playwright + ChatGPT requirements + Copilot Playwright + ChatGPT requirements + Copilot + Inline instructions Playwright + ChatGPT requirements + Copilot + Central instructions Playwright + ChatGPT requirements + Copilot + Central instructions + Audit + Stryker Applitools Autonomous Visual Testing Applitools Autonomous
  • 4.
    Use requirements togenerate tests.
  • 5.
  • 9.
    Use requirements togenerate tests.
  • 14.
    Code Coverage Measuring howmuch we’re testing
  • 15.
    Code coverage soundsuseful. It’s a trap. 15
  • 16.
    I’ll test 100%of the code. @housecor
  • 17.
    Make the toolsay 100%.
  • 18.
    When a measure becomesa target, it ceases to be a good measure. Goodhart’s Law
  • 20.
    Autonomous self-maintaining websitetesting ✓ Easy one-click setup Everything is included out of the box, from devices & browsers to test scheduling. ✓ Automatic website discovery Automatically crawls and finds all the pages in your website given its URL. ✓ Comprehensive coverage Automatically reports of any HTTP issue, content change or rendering difference ✓ Adjusts to website changes Automatically detects and reports new, missing or faulty pages on every test run.
  • 21.
    Interactive test authoringin plain English ✓ Author tests in plain English No coding or locators skills required. Easiest to understand and maintain. Interactively record test steps Speed up authoring by recording steps in English, and easily refine manually if needed. ✓ Debug tests interactively From setting breakpoints to step by step execution. Do everything a code IDE can do. ✓ 100% SaaS No need to install a local recorder. Record and playback tests exactly as they will run.
  • 22.
    Functional data driventesting Robust textual assertions Verify everything you can do with coded tests from string formats to numeric values. ✓ Test parameterization Execute tests with explicit arguments or per row in a test arguments dataset. ✓ Capture data in variables Extract value fragments and use them in subsequent test steps. ✓ Execute JS code When you need to describe complex logic, reuse a utility or manipulate the page.
  • 23.
    Visual assertions poweredby VisualAI ✓ Complete coverage Complete visual and functional coverage across multiple devices and browsers. ✓ Fewer test steps 80% less test steps to author and maintain for the entire lifetime of the test. ✓ Scalable maintenance Accept or reject differences once across all pages, devices and browsers Cutting edge computer vision Accurately mimics the human eye. 10+ dev years / 4B+ tests. Dynamic pages.
  • 24.
    API testing Initiate HTTPrequests Custom authentication, reset app state, data cleanup, or verify the backend state. ✓ Fully configurable Specify HTTP verbs, cookies, headers, and body. Use variables and parameters. ✓ Capture response details Verify response details or store it in variables to use in subsequent test steps ✓ Capture application requests Capture outbound client requests and responses to verify or use their data.
  • 25.
    LLM assisted testauthoring Automatic error correction Automatically corrects syntax and grammar mistakes ✓ Automatic step simplification Automatically splits complex steps to multiple simpler ones ✓ Automatic step disambiguation Automatically disambiguates step to clarify how the step will be executed ✓ Intent based testing preview Automatically generate test steps and test data given a test goal.
  • 26.
    LLM intent basedtesting ✓ Automatic error correction Automatically corrects syntax and grammar mistakes ✓ Automatic step simplification Automatically splits complex steps to multiple simpler ones ✓ Automatic step disambiguation Automatically disambiguates step to clarify how the step will be executed Intent based testing preview Automatically generate test steps and test data given a test goal.
  • 27.
    Summary Playwright + ChatGPT+ Copilot + Instructions • Use requirements or generate a requirements doc via LLM. • Tell the LLM to use the requirements doc to generate tests. • Audit your markup to assure it’s accessible. • Iterate. Tell the LLM about your standards (i.e. use Playwright’s recommended locators). • Work through a feature at a time to keep scope reasonable. • Use LLM instructions file (to streamline prompts and standardize output. Applitools Autonomous • Provide URL and it automatically crawls and generates. • Capture data in variables for reference later. • Call APIs and capture responses to validate expected server-side data. • Can generate, host, and administer reliable visual tests too.
  • 28.
    Want to exploreAI-powered test creation with Applitools Autonomous? Schedule a call with a Testing Specialist applitools.info/v2j Cory House | @housecor Consulting: reactjsconsulting.com Courses: pluralsight.com and dometrain.com
  • 29.
    Questions? Cory House |@housecor Consulting: reactjsconsulting.com Courses: pluralsight.com and dometrain.com