QApilot's CoWork - 3x Mobile Automation. Same QE Team.
by•
CoWork turns existing test cases into executable mobile automation with AI planning, human-approved replanning, and real-device execution on iOS, Android, and Flutter.
mobile QA is so underserved, 3x is a bold claim 👀 which platform u nail first?
Report
Congratulations
Report
The thing that kills most AI test-gen tools in practice is maintenance: the agent writes a suite, the UI changes next sprint, and now you're drowning in false failures. Does CoWork self-heal selectors when the DOM shifts, and can it tell a real regression from a benign markup change? That signal-to-noise ratio is what decides whether teams keep the suite or quietly delete it.
@mikebrandswarm - Yes, maintenance is the biggest pain for QA teams. The execution engine that executes the tests written by CoWork does AI-Self healing if the DOM shifts. We have equipped it with a bunch of fallbacks that enables it to identify where to click on when DOM shifts. The fallbacks range from text, image, parent-child hierarchies, etc., essentially solving the fragile broken tests.
Report
The honest-fail behaviour looks great. I'm curious about human-in-loop at scale though, if a nightly run throws 12 approvals at 2am, you're babysitting again. Does CoWork remember an approved change so it stops re-asking on every run after that?
@amanpreet_zop - Definitely. CoWork gives the control to the quality assurance engineer to ingest the approved changes into the test suite so that the updated replans act as the original plan for all the future runs.
Congrats on the launch! I really like the idea of turning existing test cases into something that can actually be executed without rebuilding everything from scratch.
The balance between AI automation and human approval is a smart approach. Pausing when input is needed instead of making assumptions feels much more reliable for real-world testing.
Report
I really like that this project still treats humans as a fundamental part of the QA process.
QA is extremely delicate. AI can be useful for speed and execution, but it still tends to get confused, assume things, or hallucinate its way through uncertainty.
So the human-approved replanning part makes a lot of sense to me. I’d much rather have the system pause and ask when something changes than confidently continue with a broken or misleading test.
Curious how you think about trust over time: how do QA teams review what the AI changed, and how do they decide when an approved adaptation should become part of the test suite?
Report
How does CoWork handle changes in app UI or user flows over time while keeping generated mobile automation tests reliable and minimizing maintenance?
Replies
Macaly
mobile QA is so underserved, 3x is a bold claim 👀 which platform u nail first?
Congratulations
The thing that kills most AI test-gen tools in practice is maintenance: the agent writes a suite, the UI changes next sprint, and now you're drowning in false failures. Does CoWork self-heal selectors when the DOM shifts, and can it tell a real regression from a benign markup change? That signal-to-noise ratio is what decides whether teams keep the suite or quietly delete it.
QApilot's CoWork
@mikebrandswarm - Yes, maintenance is the biggest pain for QA teams. The execution engine that executes the tests written by CoWork does AI-Self healing if the DOM shifts. We have equipped it with a bunch of fallbacks that enables it to identify where to click on when DOM shifts. The fallbacks range from text, image, parent-child hierarchies, etc., essentially solving the fragile broken tests.
The honest-fail behaviour looks great. I'm curious about human-in-loop at scale though, if a nightly run throws 12 approvals at 2am, you're babysitting again. Does CoWork remember an approved change so it stops re-asking on every run after that?
QApilot's CoWork
@amanpreet_zop - Definitely. CoWork gives the control to the quality assurance engineer to ingest the approved changes into the test suite so that the updated replans act as the original plan for all the future runs.
Upstream
Looks so useful!
ApyHub
Congrats on the launch! I really like the idea of turning existing test cases into something that can actually be executed without rebuilding everything from scratch.
The balance between AI automation and human approval is a smart approach. Pausing when input is needed instead of making assumptions feels much more reliable for real-world testing.
I really like that this project still treats humans as a fundamental part of the QA process.
QA is extremely delicate. AI can be useful for speed and execution, but it still tends to get confused, assume things, or hallucinate its way through uncertainty.
So the human-approved replanning part makes a lot of sense to me. I’d much rather have the system pause and ask when something changes than confidently continue with a broken or misleading test.
Curious how you think about trust over time: how do QA teams review what the AI changed, and how do they decide when an approved adaptation should become part of the test suite?
How does CoWork handle changes in app UI or user flows over time while keeping generated mobile automation tests reliable and minimizing maintenance?