How Protopia’s Solo QA Cut Multi-Region Deployment Testing From Three Days to One With Amikoo
Before Amikoo, Protopia's regional expansion turned every deployment into three days of manual regression work for its only QA engineer, often pulling developers in to help. With Amikoo, that same QA engineer, who had no prior test-automation experience, now runs a repo-native Playwright suite across regions, cutting deployment testing to about one day and catching three defects before production.
Maggie Hernandez, QA Engineer at Protopia, on going from zero test-automation experience to owning a multi-region Playwright suite.
Impact at a glance
Before Amikoo, I was manually testing everything. It was very time-consuming for one person to do.
The challenge
Protopia's AI-powered platform builds personalized donor experiences for alumni, donor, and student engagement teams, and connects job seekers with expert resume feedback in exchange for charitable donations.
The company ran its entire system in a single region until customers in other countries began arriving. Because the platform handles alumni, donor, and student data, privacy requirements meant those customers' systems had to live in their own regions: Asia, Australia, the UK, Canada, etc. Each one had to behave exactly like the others, which made reliable regression testing both more important and several times more work.
Every deployment then meant running the same workflow, regression, and integration tests by hand in each of them. That work landed on Maggie Hernandez, Protopia's only QA engineer, who carries production support alongside testing, and it was more than one person could absorb. For deployments, she pulled the development team in to help, which meant engineers losing days to manual regression every time Protopia shipped.
She had moved into QA from development as the company grew, and her test-automation experience at that point was none. Selenium was evaluated and set aside, its learning curve steeper than the problem it was meant to solve. Raw Playwright came next, with courses alongside it, and the tests took longer to write than the time they saved.
That was my fear: if I don't understand what I'm doing, how can I affect the system negatively? If I don't understand the agent I'm using.
How Protopia uses Amikoo
Maggie connected Amikoo to Protopia's existing GitHub repositories and described, in plain language, what the product does and how a user moves through it. Amikoo read the repositories and returned working tests before she committed anything.
- Plain-English workflows in, Playwright tests out. She describes the flow she already knows by heart. Amikoo writes the selectors and assertions.
- The tests live in Protopia's codebase. Standard Playwright TypeScript, committed to their own repositories and reviewable like any other change, not locked inside a proprietary tool.
- She corrects, rather than authors. Protopia's framework renders field names that differ from the identifiers in code, which broke the first generated tests. Resolving it was an easy exchange with the agent; there was no need for an entire rewrite.
- Started with the flows users actually take. The short form, long form, and chatbot happy paths first, with the internal approval workflow in progress.
Because of the connection between the repos or the GitHub repositories we have with Amikoo, it started working when I was explaining it.
Results
- Developers stopped testing. The manual passes that used to take engineers off the roadmap now run without them. Release windows no longer cost the development team days.
- A full multi-region release is tested in a day. Including a heavy ticket load, and by one person, whereas the same pass used to run two to three days and borrow help from the development team to finish.
- The suite covers what gates a release. 28 Playwright tests across six files in Protopia's first few months on Amikoo, covering 20% of their main user workflows. The scope is deliberate: the priority was the flows that block a deployment, not accumulating test cases, and the ones in place are doing their job.
- Three defects caught before reaching production. Bugs that previously surfaced only when someone walked the workflow by hand now surface in an automated run, in a system where the same workflows deploy across every region.
What's next for Protopia and Maggie
Protopia is a few months in, and the suite is still growing. Coverage is expanding from the happy paths into the internal approval and rejection workflow, and running Amikoo alongside Protopia's email platform is on the list, so email previews get verified inside the same automated run.
Maggie is now running Amikoo's Worker, which generates tests from every change in a pull request, so coverage grows with the code instead of waiting on a testing pass after deployment. This will soon mean that a multi-region release is tested in hours rather than a day, with most of it done before anyone asks her to start.
What made any of it work is that she can follow what Amikoo does. The tests are readable; they live in her team's repository, and when one is wrong, she says so and watches it change.
Get your developers out of manual regression.
Connect your repository, describe one workflow in plain English, and review the Playwright tests before you commit anything.
Book a demo