MerivaneBook a demo

It runs the scripts.You watch every step.

The agent drives your configured LIMS, QMS, or MES in a test account, one step at a time. Each step passes or fails with a hashed, time-stamped screenshot. When the agent is not sure, it stops and hands you the controls.

Book a demoWhere it sits in the chain
One chain, end to end
  1. Approved requirement
  2. Test scripts
  3. This pageAutonomous execution
  4. You take over
  5. Defects logged
  6. Signed OQ / PQ report
Execution

It drives your system like a tester would.

No API, no plug-in. The agent opens the browser, reads the screen, and performs each step of the approved script in a test account you control.

  • Works on the live UIAny browser-based GxP system you can give a test account to.
  • Verified against system stateEach expected result is checked on screen, not assumed.
  • Auto or manual, per stepSwitch modes at any step. The run continues from where you leave it.
OQ-2214 · Step 4 of 96LIMS-UAT · QA-02
AutoManual
lims-uat.internal/records/BR-2219
Batch record BR-2219 · Line 4In review
Sample result: 4.2 mg
Reviewer comment
Reviewed by QA-02Approve record
Screenshot capturedsha-256 · 14:32:07 UTC · locked
Evidence
Evidence

Every step leaves evidence.

A screenshot is captured at execution time for each step, hashed, and time-stamped in UTC. The hash is anchored in the audit trail. A later change to the file is detectable.

  • Captured at execution timeNot reconstructed after the run.
  • Hash-anchoredsha-256 of each file, written to an append-only trail.
  • AttributedAI steps and human steps carry the name of who acted.
OQ-2214 · Evidence log6 of 96 steps shown
StepActionResultEvidence
1Log in as QA-02Pass9f3c…a71e14:31:02 UTC
2Open batch record BR-2219Pass4b0e…c2d914:31:19 UTC
3Enter sample result 4.2 mgPassd71a…08f414:31:44 UTC
4Approve record · Priya N.Passe2c8…5b1314:33:10 UTC
5Verify audit entryPass77ad…f90c14:33:41 UTC
6Print certificateFail · DEF-0311c5f…3e6a14:34:05 UTC
Step 4 performed by a person. Same hash, same time stamp.Locked
Handover

When it is unsure, it stops.

Screens change. Dialogs appear. Sessions expire. Instead of guessing, the agent pauses, tells you what it expected and what it saw, and hands you the controls.

  • You take overYour actions are captured with the same time stamps and hashes.
  • Hand control backThe run continues from the next step.
  • RecordedEvery handover is written to the audit trail.
Step 4 needs youExpected the approval confirmation. Seen: a session time-out warning. Take over, or skip this step and continue.
Take overSkip step
Priya N. completed step 42 screenshots captured · returned control at 14:33:10 UTC
Systems in scope

Any browser-based GxP system you can give a test account to.

LIMSQMSMESERPEDCeDMSeTMFCTMS
Next in the chain

What happens with the evidence.

BeforeTest script generationFrom approved requirement to scripts with steps, expected results, and risk levels.AfterEvidence and audit trailAppend-only, hash-chained entries for every step, AI or human.AfterDefect managementFailed steps become locked defect records, assigned and tracked to closure.

Questions about execution.

If your question is not here, ask us. We reply within one working day.

Does the agent need an API or a plug-in on my system?

No. It works on the browser UI, in a test account you configure. If a person can perform the step in a browser, the agent can.

Can it run against production?

Execution runs against the environment you point it at, usually UAT or a validation instance. Access is role-based and set by your administrator.

What if a step fails?

The step is marked as failed with its evidence. A locked defect record is created, linked to the step, and the run continues or pauses according to your risk setting.

Now onboarding validation pilot partners

See a full run on one of your systems.

Bring one approved requirement. We show you the chain end to end, on a system you already use, in under an hour.

Book a demo
Reply timeOne working day
FormatLive run, 45 minutes