Software Testing with AI Agents | Agentic AI Testing

See how ai agents generate and critique stories, verify quality with acceptance criteria and definition of ready, and publish to Jira in a quality engineering workflow.

Demonstrate defining test cases from acceptance criteria using AI agents, post test cases to Jira, and design tests with industry practices and automation.

Demonstrates creating an acceptance test from a user story by automating flight booking with browser use, logging steps and data on blaze demo.com, and validating test cases.

Demonstrates automated tests with copilot in agent mode using playwright and a page object model to reuse test cases such as navigating home page and finding the cheapest flights.

Explore automatic peer review of pull requests with AI agents that fetch latest commits from GitHub, analyze code, generate review comments, and propose improvements for maintainability and testing.

Install Node.js and npm by downloading the prebuilt installer from nodejs.org, selecting your version and OS, and running the installer; then verify with node -v and npm -v.

Install Visual Studio Code and configure it for Java projects with Maven and a debugger. Import repository and use extensions like GitHub actions for continuous testing and CI in Java.

Explore what intelligence is, how ai is defined, and the core ai modules: machine learning, natural language processing, computer vision, and speech, plus the role of training models and algorithms.

Explore four core machine learning types—supervised, unsupervised, reinforcement, and deep learning—and learn how labeled data, autonomous clustering, trial-and-error feedback, and neural networks power AI and large language models.

Explore how training data drives AI performance, from extraction and ETL to transformation and load, ensuring clean, relevant, and unbiased data across images, video, text, and code.

Explore the token, a unit of text used in large language models that can be characters, words, or spaces, and how prompts are decomposed into tokens using a tokenizer.

Learn retrieval augmented generation (rag) and how encoders and a vector database empower up-to-date answers from documents and policies. See practical rag demos with ChatGPT and Flowwise AI.

Fine-tune AI models by data-driven training, prompt-based customization, and guardrails to improve task-specific performance, such as coding, while controlling output with temperature, top-k, and top-p.

Study the architecture of ai agents, from data interaction and integration to orchestration and large language model–driven reasoning, security, logging, and human-in-the-loop governance.

Configure OpenAI Codex by linking your GitHub repository, enabling multifactor authentication, and setting up an environment with dependencies, environment variables, and internet access.

Create a proper agent.md by defining instructions, config, pages, tests, and test data to ensure the agent runs only when instructed. Explain why a clear agents md matters.

Use Codex and an agent to generate and expand test data, including both valid and invalid samples, and manage changes via branches and pull requests.

Showcases using an agent and Open AI Codex to generate negative tests with invalid and empty data, confirming failures without executing tests, and guiding master branch workflows.

Explore browser use, a free, open-source wrapper that lets an agent control a browser with vision and HTML extraction to perform tasks across multiple tabs, using multiple large language models.

Resolve browser errors by understanding how Playwright installs the latest browser versions, managing dependencies, and adjusting the browser folder names to match expected versions during tests.

Configures and runs a browser agent to compare ChatGPT pricing with deep seq v3, using a Python model and Chrome driver with environment variables.

Agentic AI testing shows how to generate end-to-end functional test cases for a travel site using browser-based exploration, element identification, and page object model outputs.

Discover Microsoft Autogen, an open-source multi-agent framework with a drag-and-drop studio for orchestrating agent conversations in dotnet or Python, focusing on Autogen 0.2.

Install Microsoft Autogen in an isolated VM, use the correct pip package (autogen) and OpenAI dependencies, and verify by running a simple chat bot.

Define an agent to chat with a large language model, building a simple chatbot with Autogen or an OpenAI API client, leveraging built-in memory and environment variables.

Autogen enables two agents to converse and critique a Python function for square roots, iterating back-and-forth debates to converge on a robust solution with unit tests and caching.

Explore autogen workflows that empower two agents to iteratively craft and critique user story-based test cases, guided by acceptance criteria and prompts that push final definitions to Jira-ready outputs.

Activate Copilot to build an AI agent with GitHub Copilot. Install the Playwright extension, sign in with GitHub in Visual Studio Code, and enjoy 2000 free code completions.

Learn to install the Playwright MCP server, choosing global or local setup, start and stop the server, and configure an MCP.json for local use with npm and VSCode.

Validate the Playwright MCP server install by using Visual Studio Code in agent mode to automate a flight search and Copilot-assisted test generation.

Install flow wise locally by following official docs, ensure node.js and npm are installed, and use GitHub readme for commands; run on localhost:3000 and manage versions with npm updates.

Build an ai agent using prompt chaining with a supervisor orchestrating worker nodes and a chat model, connected via an OpenAI api key to generate story prompts and titles.

Read a user story from a file, generate test cases with agent one, then agent two creates BDD scenarios and saves them to disk to automate tests.

Generate automated web tests from a webpage using ai agents, fire crawl, a crawler, and retrieval augmented generation to create executable playwright scripts.

Fix a flaky automated front-end test with Playwright by inspecting selectors, updating test data and expected strings, and planning a refactor toward a page object model.

Extend an end-to-end flight search test case, use Playwright with ChatGPT to generate a first draft JavaScript test, and manage test scripts with git.

Learn to configure Playwright with a comprehensive config file—defining test locations, match patterns, timeouts, retries, parallel workers, browser targets, reporters, and outputs.

Configure multi-browser testing with a JSON project file, run tests sequentially across Chrome, Firefox, and Edge in headless or headed modes, and emulate browsers using a Chromium driver.

Explore zero step, a tool that wraps ChatGPT to remove element selectors and verify headers with plain-english prompts in Playwright tests.