Axl Test · in pilotStatus: Pilot

Test your Ivalua build against its design

Axl Test turns the design package into test scenarios, runs them against your Ivalua environment and shows each place where the environment and the design differ.

Run Axl

What it does today

  • Writes scenarios from the design — Axl authors test scenarios from your design package: requirements, user stories and the field dictionary. Each one is written as Given/When/Then steps that people on the project can read and challenge.
  • Proves each scenario can fail — A scenario can't be approved until it has been seen failing. That keeps tests that always pass out of your results.
  • Tests the build as it lands — Some checks only look at screens and never change data. Others are written before the build and show what is still missing. Tests that create records run only where you allow them.
  • Keeps production read-only — Axl Test changes data only on environments you name for it; production stays read-only.
  • Reports by requirement — Each run produces a PDF and an HTML report: what ran, what passed, what failed, what is left for manual testing, and each difference between the design and the environment.
  • Drafts the defects — Each failed scenario gets a defect draft formatted for Jira, Azure DevOps or Ivalua Extranet. Someone on the project reviews and files it.

How a pilot runs

  1. We agree the scope with you: one module or business process, its design package and the environments Axl may use.
  2. Your team or Modali sets up the environments in Accelerate and stores a test account for each one. The password is stored encrypted; Accelerate can't read it, and it is used only for runs on that environment.
  3. Axl writes the scenarios from the design. A test lead reviews them, and only scenarios that have been seen failing can be approved.
  4. A test lead starts a run. A Modali operator approves it, and then it runs against your environment.
  5. The run page shows the verdict, the results by requirement, the PDF report and the defect drafts.

Verdicts you can act on

Every run ends with one of three verdicts. Pass means the environment matches the design for everything that ran. Hold means something needs attention: the environment differs from the design, a test is wrong and Modali fixes it, or a safety rule stopped the run. Error means there was no verdict, for example because an account was missing or the run wasn't approved in time.

Some Ivalua records can't be deleted once they are submitted, so we agree with you up front how test records are created and handled.

Coverage that doesn't flatter

Requirement coverage counts approved suites only. Draft suites are shown separately and labelled. Scenarios that need a person are counted as manual, not folded into the automated figure, so you can see what is still on your testers' list.

Who can do what

Access follows the Accelerate project. A test lead reviews scenarios and starts runs. A test admin manages environments, test accounts and pipeline tokens. People from the client organization can, by default, read everything and request new scenarios, but can't start runs or approve them. Every change to access is recorded in the audit log.

Starting runs from a pipeline

A test admin can create a token that lets a deployment pipeline request a run for each release drop. The token is limited to the environments, suites and modes you choose (read-only unless you need more), is shown once, and can be revoked at any time.

The run still waits for a Modali operator's approval. The pipeline step waits for the verdict and passes on the same result as the run page. If the run isn't approved in time, the step ends without a verdict instead of passing. Sending the same drop twice starts one run.

Proof of concept, September 2026

Modali ran Axl's test agent against the intake module of a US state e-procurement platform. Axl planned 64 scenarios from the design package and field dictionary. 36 ran automatically and 21 passed. 15 showed differences from the design, across 5 of the 30 requirements. The remaining 28 were listed for manual testing and reported separately. A later read-only run took 2 minutes 38 seconds.

The design was provisional and the environment was not yet configured to it, so the team treated the 15 results as differences to review, not defects.

Pick one module and one release. We'll set up a pilot and show you the first report.

Run Axl