I’ve been working in software for more than 10 years. And I’ve always been obsessed with making life smoother and easier both in personal routines and working with my fav QA and Engineering teams. After reading Atomic Habits (which I really recommend to everyone who haven’t read yet), my life’s credo became: Optimize it all – from daily routines to big projects. That’s how you achieve the biggest success. Optimize it all – from daily routines to big projects. That’s how you achieve the biggest success. Optimize it all – from daily routines to big projects. That’s how you achieve the biggest success. I wouldn’t say I lived differently before reading the book, but it confirmed that my way of thinking was on the right way. Optimization has always been my strength and what I really enjoy. When AI started entering our lives and while some people worried about layoffs, I started changing my mindset and I realized AI could give me even more power. As a QA Manager leading a team of 7 talented QAs, I’ve always believed that process optimization is the backbone of quality assurance. Over the past year, I’ve actively combined my management experience with AI solutions and multi-step agents to build workflows that successfully improved my team’s efficiency. The Goal: Kill the Manual Test Case Grind The Goal: If you’ve ever set up a Jira API integration, you know it’s not exactly “click and done”. For me, it was a mix of OAuth headaches, endless API docs, and too many “why is this not returning anything?” but not at a polite manner. But once I got it working, the payoff was huge! My team now has a GPT-powered Structured Test Case Generator that transforms Jira tickets into ready to publish, standardized test cases in seconds. Before this project, test case creation looked like this: Before this project, test case creation looked like this: Before this project, test case creation looked like this: Open Jira, find the ticket.
Read through the user story, acceptance criteria, and sometimes even comments.
Read Confluence.
Manually write the test case in our preferred structure we approved with the QA Team.
Copy it into our test management tool. We use Testomat.io, which I really recommend (cheaper than TestRail and has a solid, fast functionality). Open Jira, find the ticket. Read through the user story, acceptance criteria, and sometimes even comments. Read Confluence. Manually write the test case in our preferred structure we approved with the QA Team. Copy it into our test management tool. We use Testomat.io, which I really recommend (cheaper than TestRail and has a solid, fast functionality). Testomat.io All these actions were repetitive, slow, and prone to inconsistencies. I wanted an AI agent that could: I wanted an AI agent that could: I wanted an AI agent that could: Read a Jira ticket directly via API.
Extract the summary, description, and acceptance criteria, and look at Confluence.
Generate structured test cases in our exact QA format.
Offer edits, add negative test cases, and merge cases before publishing.
Push them straight to Testomat.io once approved. Read a Jira ticket directly via API. Extract the summary, description, and acceptance criteria, and look at Confluence. Generate structured test cases in our exact QA format. Offer edits, add negative test cases, and merge cases before publishing. Push them straight to Testomat.io once approved. Testomat.io The Build:  GPT + Jira API + Testomat.io The Build: It took me a few sleepless nights, but finally I created the Structured Test Case Generator in GPT with: Model: GPT-4o  (fast + accurate).
Capabilities: Web search, Canvas, Image Generation, Code Interpreter for formatting/validation.
Actions: api.atlassian.com → Jira ticket retrieval app.testomat.io → publishing test cases. Model: GPT-4o  (fast + accurate). Capabilities: Web search, Canvas, Image Generation, Code Interpreter for formatting/validation. Actions: api.atlassian.com → Jira ticket retrieval app.testomat.io → publishing test cases. What it does: What it does: Accepts a Jira ticket ID like TEST-123.
Calls fetchJiraTicket to get details and look at Confluence.
Generates structured test cases with: Title Preconditions Steps to Reproduce Expected Results or whatever template you share. Accepts a Jira ticket ID like TEST-123. Calls fetchJiraTicket to get details and look at Confluence. Generates structured test cases with: Title Preconditions Steps to Reproduce Expected Results or whatever template you share. Title Preconditions Steps to Reproduce Expected Results Aaaand Prompts the user to review, edit, or add more cases.
Publishes to Testomat.io if approved.
Even can merge and optimize steps. Prompts the user to review, edit, or add more cases. Publishes to Testomat.io if approved. Even can merge and optimize steps. Here's a schema process of how it works: The Jira API Challenge The Jira API Challenge Getting GPT to talk to Jira wasn’t just a copy-paste job. My main roadblocks were: OAuth 2.0 Setup – aligning the callback.
URL with GPT’s environment took trial and error. It was freaking hard to resolve.
Permissions – one missing API scope = hours of confusion.
Data Formatting – Jira’s JSON is… verbose. I had to filter just the essentials.
Edge Cases – missing fields or messy acceptance criteria meant adding fallbacks to avoid bad outputs. OAuth 2.0 Setup – aligning the callback. URL with GPT’s environment took trial and error. It was freaking hard to resolve. Permissions – one missing API scope = hours of confusion. Data Formatting – Jira’s JSON is… verbose. I had to filter just the essentials. Edge Cases – missing fields or messy acceptance criteria meant adding fallbacks to avoid bad outputs. But in a result we actually won But in a result we actually won Before we spent: 45-60 minutes to manually write & format test cases for a complex ticket. After: 3-5 minutes to review AI-generated cases, polish if needed, and submit “Publish” to Testomat.io. Also, other wins we’ve: Also, other wins we’ve: Reduced test case writing time by ~50%.
Standardized all formats across the team (it was challanging a lot and a pain point when you work with other QAs).
Increased coverage with negative and edge cases.
Eliminated copy&paste fatigue with direct Testomat.io integration. Reduced test case writing time by ~50%. Standardized all formats across the team (it was challanging a lot and a pain point when you work with other QAs). Increased coverage with negative and edge cases. Eliminated copy&paste fatigue with direct Testomat.io integration. Example Flow Example Flow Then the agent asks: “Would you like to edit, add negatives, merge, or publish?” Once you  click publish → it’s in Testomat.io, ready for execution. Lessons Learned Lessons Learned APIs are the glue – but OAuth setup inside GPT requires patience. Prompt clarity = output quality – be specific about formatting. Integration matters – direct publishing to Testomat.io is the real game-changer.
Life hack: In addition to integration to Cnfluence, add any project-related test docs to the Agent’s Knowledge section to avoid irrelevant or generic test cases. APIs are the glue – but OAuth setup inside GPT requires patience. Prompt clarity = output quality – be specific about formatting. Integration matters – direct publishing to Testomat.io is the real game-changer. Life hack: In addition to integration to Cnfluence, add any project-related test docs to the Agent’s Knowledge section to avoid irrelevant or generic test cases. Below as a bonus you can find the detailed technical step by step map of creating GPT agent. Hopefully it can help for the future optimizations of the daily routine. Final Thoughts Final Thoughts This project proved something important: AI isn’t replacing QA engineers – it’s making us faster, more consistent, and more focused on what matters most: testing quality. If you’re a QA lead or engineer, try building your own GPT agent + API workflow. For sure, the setup might test your patience (hello, Jira OAuth =)), but the payoff is worth it. Your turn: Your turn: Have you tried connecting GPT to Jira or your own test management tool? What challenges did you face and what wins did it bring?

Walkthroughs, tutorials, guides, and tips. This story will teach you how to do something new or how to do something better.

6 Leadership Mistakes That Made Me Better (I hope 😅)

I Replaced Test Case Busywork With a GPT Agent

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

6 Leadership Mistakes That Made Me Better (I hope 😅)

6 Leadership Mistakes That Made Me Better (I hope 😅)

When to Automate Testing and When to Stay Manual

The Definitive 227 Item Quality Assurance Checklist to Test Before Launching New Software

The Noonification: How Often Do NFTs Pass The Howey Test? (1/13/2023)

Darwin's Hybrid Intelligence to Align AI & Human Goals for Startups & VCs

6 Leadership Mistakes That Made Me Better (I hope 😅)

6 Leadership Mistakes That Made Me Better (I hope 😅)

When to Automate Testing and When to Stay Manual

The Definitive 227 Item Quality Assurance Checklist to Test Before Launching New Software

The Noonification: How Often Do NFTs Pass The Howey Test? (1/13/2023)

Darwin's Hybrid Intelligence to Align AI & Human Goals for Startups & VCs

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps