Regression testing

Ship the change.
Keep everything else working.

Regression testing catches what your new code quietly broke. Bug0 runs it on every pull request, heals it when your UI moves, and a forward-deployed engineer checks every failure before it reaches you.

Built on Passmark, our open-source Playwright library for AI regression testing.

Updated Jun 14, 2026

  1. It worked last week

    You merge a small fix on Friday. Monday, a customer emails: checkout is broken. The code you touched was nowhere near checkout.

    That is a regression. A feature that worked before your change, and stopped working after it. Regressions are the tax you pay for shipping fast, and they get more expensive the faster you ship.

  2. Bug0 regression testing running on a pull request, self-healing when the UI changes and verified by a forward-deployed engineer

  3. What is regression testing?

    Definition

    Regression testing is re-running tests after a code change to confirm that features which already worked still work. It does not check new behavior. It checks that old behavior did not break. Teams run it after every change, refactor, dependency bump, and release.

    The name comes from the bug behavior: the software regressed to a broken state. Good regression coverage turns “I hope nothing broke” into a green check you can trust.

  4. When you need it

    Run regression tests any time the code or its dependencies move. In practice that means after every pull request, during refactors, when you upgrade a library or framework, and before each release. On an agile team shipping daily, regression is not an event. It is something that has to happen on every merge.

  5. Regression vs functional, integration, and smoke testing

    These terms overlap, which is why they get confused. The difference is timing, not technique.

    TypeWhat it checksWhen you run it
    Regression testingThat features which worked before a change still workAfter every change
    Functional testingThat a feature behaves the way it was specifiedWhile building or changing that feature
    Integration testingThat separate modules and services work togetherWhen systems connect
    Smoke testingThat a build is stable enough to test furtherRight after a deploy, before deeper tests

    Your functional and integration tests become your regression suite the moment you re-run them on every change. Regression is the job. The other tests are how you do it.

  6. Why regression suites rot

    Most teams do not lack regression tests. They lack a regression suite they trust. Three things break it over time:

    • 01Brittle selectors. A test keyed to .submit-btn breaks the day a developer renames the class, even though the feature still works.
    • 02Maintenance debt. Teams spend roughly 40% of their QA time fixing tests instead of writing them. The suite becomes a second codebase nobody wants to own.
    • 03Flakes that hide real bugs. Once a suite goes red for no reason often enough, the team stops reading the red. Then a genuine regression slips through behind the noise.

  7. How Bug0 runs regression testing

    Bug0 is Passmark as a service. Your forward-deployed engineer builds the suite on the engine, and the engine keeps it running while your app changes.

    • Plain-English steps, real Playwright actions. Your engineer writes steps like “add the plan to the cart and check out.” Passmark resolves each one into a Playwright action at run time, so there are no hand-keyed selectors to break.
    • Self-healing on UI change. When a cached step fails because the page moved, Passmark re-resolves it with fresh AI and caches the new path. A renamed button does not turn the suite red.
    • Assertions checked by two models. Each assertion runs on Claude and Gemini in parallel, and a third model settles any disagreement. That consensus is what keeps false failures out of your pull requests.
    • Verified by a human, gated in CI. The suite runs on every pull request. Your engineer reviews each failure, decides bug or flake, and files real bugs with video and repro before the change ships.
    github.com/acme/app/pull/847

  8. A service, not another tool

    Most regression testing tools hand you a platform and leave the work with you: you still write the tests, chase the flakes, and own the maintenance. Bug0 is the opposite. You give us staging access. We build, run, heal, and verify the suite, and report results where your team already works.

    The engine underneath, Passmark, is open source, so there is no lock-in. The outcome is a regression suite that stays green without taking a single engineering hour from your team.

“Bug0 is the AI QA platform behind Prospyr, the practice management software for aesthetics clinics. It tests our web app continuously and points us to exactly where the app broke when something fails. We catch issues early instead of in production, which keeps a HIPAA-compliant product stable.”
Portrait of Greg KopyltsovGreg KopyltsovCo-founder and CTO, Prospyr
1-2 weeks
100% of critical flows covered
0%
Flake rate, tests auto-heal
$150K+
Saved vs an in-house QA hire

Stop owning the regression suite.

Give Bug0 staging access. Your forward-deployed engineer builds the regression suite on Passmark, runs it on every pull request, and keeps it green as your UI changes. You ship. We catch what broke.

Managed AI QA with a forward-deployed engineer from $2,500/mo. Results in your first week.

Frequently asked questions

On every change. A regression suite is only useful if it runs before code merges, not on a quarterly cycle. Bug0 runs your suite on every pull request, so a regression is caught while the change is still in review.
Yes, and it should be. Manual regression does not scale past a handful of flows. Bug0 automates the suite on Passmark, our open-source AI engine, then a forward-deployed engineer verifies every failure so you get real bugs, not noise.
Retesting confirms a specific bug fix works. Regression testing confirms that fix did not break anything else. You retest the one thing you changed. You regression test everything around it.
Bug0 does. When your UI changes, Passmark re-resolves the affected steps instead of failing on a dead selector, and your forward-deployed engineer handles anything the engine cannot heal on its own. Your team writes nothing and maintains nothing.
From $2,500 per month, flat. That covers the engineer, the AI engine, the infrastructure, and unlimited runs. There is no per-test or per-seat billing.

Go on vacation. Bug0 never sleeps. The AI tests every commit, every deploy, every schedule. Your forward-deployed engineer reviews every failure and files the bugs. Coverage holds while you're off the grid.

Go on vacation.
Bug0 never sleeps.

The AI tests every commit, every deploy, every schedule. Your forward-deployed engineer reviews every failure and files the bugs. Coverage holds while you're off the grid.