That's literally what QA is for. I'm not here to break my own code on purpose — that's your job. If developers had to test everything they ship, we'd never release. The boundary exists for a reason. I build it. You test it.
Third sprint in a row, same story. Devs done, you're still testing. I get it — there's a lot to cover. But the answer is obvious: we're understaffed on QA. Two more people and this queue clears. It's a resourcing problem, not a process problem.
So you found a 2px offset on a modal that 0.3% of users ever open. And now you're holding the release. The release that was supposed to go out Monday. Do you understand what that sprint costs? Do you know what the stakeholders are going to say? This is why developers don't respect QA.
I don't understand what's taking so long. The feature is done. Open the ticket, write the steps, run through it. You did exactly this last month in a day. What is "test design" even? Just test it the same way you always do. It's the same application.
There's no time to go through acceptance criteria right now. Look at the mockup. Use common sense for the edge cases. The spec is in Confluence somewhere — I wrote it last quarter. Demo is Thursday. Just test what's there and we'll clean up in the next sprint.
We moved to Scrum. There's no QA column on the board. No QA ceremony. No QA role in the framework. The team owns quality together. Which means — and I'm saying this positively — it's part of everyone's velocity now. Including yours. Which is already accounted for in the sprint.
You booked a retro slot to talk about "shift left". You came to planning to ask about "testability". You sent a Confluence doc about "quality gates". The architects already made their decisions. The sprint started three weeks ago. Test the thing when it's ready. That is the job.
Look, I'd walk you through the logic but it'd take an hour and I don't have that right now. Just trust me — the behavior is correct. If the ACs say otherwise, the ACs are wrong. That's a product problem, not an engineering problem. Mark it "by design" and let's move on.
I've been looking at what Copilot can do. It generates test cases. Real ones. And it doesn't take lunch, doesn't need two sprints to understand the architecture, doesn't ask for ACs to be clarified. I'm not saying we fire everyone right now. But when the next QA person leaves... maybe we see how AI handles it first. Just thinking out loud.
Read about real Senior QA responsibilities and activities.
ChatGPT can write tests. It can't tell you what to test. You'll find out which one matters.
