In my day-to-day work as a QA Automation Engineer, tools like Claude and Codex have become less of a “nice-to-have” and more like a second brain. Here are the top 5 ways I actually use them in real projects, with practical examples.

Generating Test Specification Documents in Minutes

Writing test specs used to take hours—especially when translating requirements from Jira into structured test scenarios.

Now, I feed the requirement directly into Codex and get a clean first draft.

Example input:

Generate a test specification for a login feature with:

• valid login

• invalid password

• locked account

• session timeout

Output (refined):

Test Case ID Preconditions Steps Expected Results

Instead of starting from scratch, I just review and refine.

👉 Result: ~70% time saved on documentation.

Bulk Code Changes Without Losing My Mind

Refactoring test code across multiple files is painful—especially when patterns change.

Using OpenAI Codex, I can describe the change once and apply it everywhere.

Example:

Update all Selenium locators from XPath to CSS selectors

Or:

Replace time.sleep() with explicit waits across test files

Instead of manually editing 20+ files, I:

Generate the updated pattern Apply it across the repo (via scripts or IDE tools)

👉 Result: Faster refactoring + fewer human errors.

Writing Python Test Scripts from Plain English

This is probably the biggest daily win.

I describe a test flow, and Codex generates a working script in Python.

Example prompt:

Write a Selenium test in Python:

1. Open login page

2. Enter username/password

3. Click login

4. Verify dashboard is visible

Generated output (simplified):

from selenium import webdriver from selenium.webdriver.common.by import By

driver = webdriver.Chrome() driver.get("https://example.com/login")

driver.find_element(By.ID, "username").send_keys("testuser") driver.find_element(By.ID, "password").send_keys("password") driver.find_element(By.ID, "login").click()

assert "Dashboard" in driver.page_source driver.quit()

I still review it (always), but it removes the “blank page problem.”

👉 Result: Faster test creation, especially for repetitive flows.

1. Research & Deep Dives Using NotebookLM

When I need to understand a new tool, framework, or testing strategy, I use NotebookLM.

Instead of:

Reading 10 different blogs Piecing together info

I:

Upload docs / links Ask targeted questions

Example:

Summarize best practices for API test automation using Python

It gives:

Structured insights Key patterns Simplified explanations

👉 Result: Faster learning with less noise.

Generating Edge Cases & Test Ideas

This one is underrated.

AI is great at thinking of scenarios you might miss.

Using Claude, I ask:

List edge cases for a payment system

Output includes:

Network failures Duplicate transactions Currency mismatches Timeout scenarios

This helps strengthen test coverage beyond “happy paths.”

👉 Result: Better quality tests with minimal extra effort.

What Still Needs Human Judgment

Let’s be real—AI isn’t perfect.

Things I never fully trust AI with:

Final test logic validation Business-critical edge cases Debugging flaky tests

AI helps you move faster—but you’re still the quality gate.

Final Thoughts

The real advantage isn’t just using AI—it’s knowing where it actually saves time.

For me, that’s:

Documentation Boilerplate code Refactoring Research Idea generation

If you’re in QA and not using AI like this yet, you’re honestly leaving a lot of efficiency on the table.