FAQ
Frequently asked questions
Clear answers about wallet credit, usage, subscriptions, and how Tycoon charges for work.
Can AI really replace a human QA engineer?
For execution, yes. For policy, no. The AI QA Engineer runs tests, reproduces bugs, writes coverage, and triages the queue at a quality that matches a mid-level human QA engineer and exceeds most solo-founder QA efforts (which is typically zero). What it does not do is decide what quality bar the product should hold to, which risks are worth taking, or what customer experience defines "shippable." Those are CTO and founder decisions. The AI does the work; humans hold the taste.
How does it know what to test for new features?
It reads the spec (Linear ticket, RFC, design doc), the related code paths, and past test history for adjacent features. It generates a test plan draft that the engineer reviews before coding starts — this is the key inversion: tests are written with the feature, not bolted on after. Most QA failures at small companies come from after-the-fact testing; the AI QA Engineer makes it cheap enough to test first. Over time it learns the patterns of your product (what always breaks, what users always misuse) and the coverage gets stronger.
What about manual exploratory testing?
This is the honest weakness. The AI QA Engineer is strong on scripted testing, regression, and API fuzzing. It is weaker on exploratory "tries to use the product the way a confused human would." For products where subtle UX issues matter (onboarding, checkout, mobile) a human exploratory session once per release is still valuable. Most solo founders run a 20-minute exploratory pass before each release themselves; the AI handles everything else. This is genuinely 95/5 where the 5% of human time is high-signal.
Does it work with my existing test framework?
Yes. Playwright, Cypress, Puppeteer, Jest, Vitest, Mocha, Pytest, RSpec, and the major frameworks are first-class. You do not migrate anything — the AI QA Engineer reads your existing test setup, contributes to it, and extends coverage from where you are. For projects with zero tests today, it starts with Playwright smoke tests on the critical paths (signup, checkout, core user flow) and expands from there at whatever pace you fund.
How does it handle flaky tests?
Flaky tests are tracked as a specific metric. Each week the AI QA Engineer ranks the top flakes, investigates root cause (timing, state leakage, env instability), proposes a fix, and either fixes it or quarantines it with a deadline. Tests that remain flaky past two weeks get retired. This loop matters because flaky tests are the number one cause of test suites becoming worthless at small companies — someone has to own the hygiene, and no human QA lead has the time. The AI does.