For QA directors and heads of quality

Fewer people, more code, and a release review on Friday.

Your team is thinner than it was, engineering is shipping more with AI, and you are still the one asked whether it is covered. Hiring and retraining cycles are slower than the releases. This is how CertiQA helps you keep control of what is tested without reading every script.

See the problems we solveRequest early access

What this looks like day to day

Four problems that do not fit in one headcount plan.

They share a root: what is tested lives in code only a few people can read.

How do we keep coverage with fewer people?

Scripts take specialists to write and maintain. When the team shrinks, the suites nobody has time to fix are the first to be ignored.

Who reviews what the AI generated?

Code and tests are produced faster than your seniors can read them. Either review becomes a bottleneck or it quietly stops.

How do we reskill without a hiring cycle?

Your manual testers hold the product knowledge. Asking them to learn a framework first delays the coverage you need now.

What do I bring to the release review?

Green or red is not enough. You want what changed since the last release, what is linked to a requirement and something a reviewer can check.

How CertiQA helps

Each problem has a specific answer.

Make tests cheaper to build and to keep

A test case is a list of plain steps on a visual canvas, so testers and domain experts can build and edit them without a framework. Shared sequences such as login become CertiBlocks: edit one and every case, suite and agent test that embeds it picks up the change on its next run. A case that belongs to several suites is one definition, not copies.

How reusable blocks work

What changes

  • Fixes are made once, not per case.
  • More people can contribute coverage.
  • Seniors maintain shared blocks, not every script.

Put a person between AI output and your suites

Assistants such as Cursor and Claude Code can draft projects, suites and cases over MCP, where each case is validated and reviewed as plain steps before joining a suite. Each project also has a change review policy. When it is on, changes to cases, CertiBlocks and suite composition wait in the Pull Request Inbox. Reviewers press View Diff, comment, then Approve and Apply or Reject. Authors cannot approve their own change.

The policy is per project and can be off, applied only to changes made with an API token, or strict. Choose it before you connect an assistant, because the protection depends on the setting.

In the product

  • A diff of each pending change.
  • Review comments on the change.
  • Admins can be exempt from review.

Reskill the people you already have

Your testers know the product, and plain steps let them add automated coverage from their first week. They can record a web flow with the browser extension, and the same step vocabulary covers web UI, API and AI agent tests. Anyone who wants to learn code can export a web UI test to Playwright TypeScript and read it next to their steps.

What this looks like for a QA engineer

A way to start

  • Choose one release-critical flow or one AI feature.
  • Have a tester build the cases.
  • Review the steps together.
  • Compare with how you test it today.

Bring more than a pass rate to the release review

Run Comparison lines up a baseline run against a current run and lists regressions, improvements and persistent failures, with the change in duration and pass rate and the error and screenshot for each case. The requirement coverage matrix shows which requirements are linked to a test.

Coverage means a requirement is linked to a case, not that the case is thorough.

One case, three surfaces

If your company ships an AI agent, the same team tests it with the same steps. See AI agent testing.

When someone asks for proof

Hand a reviewer a Release Report.

Signing off a run in the app generates a Release Report for it, with the approver and decision recorded. A project-wide export for a date range is available through the API. Either way it collects what CertiQA measured into a single export with a SHA-256 hash, and a public verify endpoint lets anyone check it was not changed.

It is evidence for a human reviewer, not a certification. The audit-log section covers the whole platform, not only the project.

What managers ask to see

FAQ

Questions QA leaders ask

How can a smaller QA team keep test coverage up?

By making tests cheaper to create and easier to review. Testers and domain experts build tests as plain steps, shared sequences such as login are built once as CertiBlocks and reused, and AI assistants can draft cases over MCP for people to review. Run Comparison shows what changed between two runs without anyone reading raw logs.

How do we review tests that AI wrote?

Open the case and read the steps. Each says what is done and what must be true, in language a QA lead, product owner or auditor can follow. For stricter control, each project has a change review policy. When it is on, edits to test cases, CertiBlocks and suite composition wait in a Pull Request Inbox as a diff, and someone other than the author approves them before they apply. The policy can be off, applied only to changes made with an API token (the kind an assistant over MCP uses), or strict for everyone except project and system admins.

Do my manual testers need to learn to code?

Not to be productive. Tests are plain steps on a visual canvas, or drafted by connected AI assistants like Cursor or Claude Code over MCP. A person edits and approves steps before any cases are run. People who want to learn code can export web UI tests to Playwright TypeScript and read it alongside their steps.

How QA engineers grow into automation
Can AI assistants write our test suites?

They can draft them. Cursor, Claude Code and Claude Desktop can create projects, suites and cases through the CertiQA MCP server, and cases arrive as plain steps. Whether those cases wait for a reviewer before they apply is up to the project's change review policy, so decide that before you connect an assistant.

AI assistants over MCP
We are shipping AI agents. Can the same team test them?

Yes. Agent tests use the same plain steps: what the user says, which tool responses are mocked and what the reply must or must not contain. CertiQA scores replies with an LLM judge, can inject tool failures and runs OWASP LLM Top 10 attack scenarios.

AI agent testing
What evidence can I bring to a release review?

Run Comparison for what regressed or recovered between two runs, the requirement coverage matrix for what is linked to a test, and a Release Report for a signed-off run or a date range. A Release Report covers run outcomes with pass rates and Wilson 95% intervals, fault-injection recovery, security posture, judge calibration, traced tool calls, accessibility findings and a platform-wide audit-log excerpt. It carries a SHA-256 hash, and a public verify endpoint lets anyone check it was not edited. It is evidence for a reviewer, not an audit or a certification.

The brief for managers
How do I see which requirements have no test?

Add requirements to the project and link cases to the requirements they cover. The project then shows a matrix of each requirement and the cases linked to it, and highlights requirements with no linked scenario. Covered means linked to at least one case, not that the case is thorough.

Coverage for automation leads
How do we try it with a team?

Request early access. Once your spot opens, start the 14-day trial, which includes the Team Pro Plus features and 500 credits, with no credit card. Pick one release-critical flow or one agent, build it together and compare the results with your current process.

See pricing

Try it on one flow with your team.

Build a suite together, decide the review policy and see what your people can read and approve. Automation leads may also want the coverage view.

Request early accessBrief for managers