TestEdge AI
English EN

Language

Sign in Request a demo
Menu
Sign in Request a demo

From requirements to visual, executable and measurable tests.

Design and record, validate, run inside your network and certify with evidence. TestEdge AI is the quality platform for teams that test banking, payments and other regulated portals and APIs.

Request a demo See how it works

The agent is installed inside your network and only opens outbound connections. No ports to open, no environments to expose.

Where each piece runs The TestEdge platform keeps the design, approvals, signatures and results. Inside your network, the agent opens the connection to the platform over outbound HTTPS, gets the signed job back as the response, runs it against your portal and your API with the secrets in your vault, and sends the verdict back the same way. TestEdge platform Design and approval Signed versions Results, coverage and gate Outbound HTTPS no inbound ports Your network Agent checks signatures Web portal API Secrets your vault
Your addresses, credentials and test data stay inside your network. The platform receives references and results. The agent always opens the connection; the job comes back as the response.

From requirement to result

Every test follows the same path, and every step leaves a trail. A case is not a test point, and a test point is not an attempt: retries never inflate coverage.

  1. Quality

    Requirement

    What the business needs to verify, with its criticality. Import it from Excel or CSV, or write it in the platform.

  2. Automation Studio

    Case and flow

    The case says what is tested; the flow, how. Draw it in Automation Studio or record it by walking through the portal.

  3. Approval

    Approved version

    Someone else reviews and approves it. Once published, it is signed and never changes.

  4. Execution Grid

    Run inside your network

    The agent picks up the job, verifies the signatures and runs every data row against your environment.

  5. Release

    Evidence and gate

    Every step with expected versus received, coverage per requirement and the release gate. Releasing is still a person's decision.

What it looks like

Real screens from the portal, with the demo's made-up data. The portal is in Spanish for now.

Automation Studio with the SALDO_CONSULTA flow on the canvas: start, get the balance, check that it answers 200, check that the balance is there, and two ends, pass and fail. On the right, the properties of the selected check.View full size
Design the flow. Each block is a call or a check, and you can see where the test goes when a check holds and where it goes when it fails.
Result of the TRANSF_SIN_FONDOS test: passed. Each step shows what was expected and what came back, for example 422 expected and 422 received, and the SHA-256 fingerprint of the evidence that stayed on the agent.View full size
Review the evidence. Every step with what was expected and what came back. The request and the response stay on the agent; the platform gets their SHA-256 fingerprint.
The demo's Release 2026.11: the gate does not pass. One criterion is met and two are not, with the cases holding them back, two warnings and the figures for design, execution, requirements met, success rate and weighted risk.View full size
Decide on a release. The gate shows which criterion is not met and which cases are holding it back. Releasing is still a person's call.
Quality summary: two requests waiting for a decision, two releases that do not pass the gate and why, 87% of runs passed in the last 7 days and 0% of releases pass the gate.View full size
Keep track of quality. What is waiting for your decision, which releases are not passing and how this week's runs are going.

Four modules, one platform

Use them together or one at a time, depending on where your quality team is today.

Quality Management
Requirements, cases, suites and plans. Frozen releases with their coverage, their gate and a report you can download as JSON or CSV.
Automation Studio
A visual flow editor for browser and API tests, with no code to write. An operations catalog from OpenAPI or Postman, versioned test data and step-by-step debugging on the canvas.
Execution Grid
Agents inside your network, environments and concurrent runs. Every run shows the path the test took through the flow, block by block.
AI Quality
Proposes cases from your requirements and helps design the flow. What it proposes is a draft: your team reviews and approves it.

What stays inside your network

TestEdge is built for teams that cannot move their credentials or their environments to the cloud.

Stays inside your network Reaches the platform
Secrets. The agent reads them from your local vault.A reference, such as vault:test-user
The real addresses of your portals and APIs.An alias per environment, such as portal.transfers
What is typed while recording, cookies and screenshots.Steps with opaque references. A batch with data in clear text is rejected as a whole.
The content of the evidence, unless your organization and the agent allow it.The verdict, a summary of each step and the SHA-256 fingerprint of every file.

  • Nothing runs without approval

    Only a published version runs, approved by a QA Lead other than its author.

  • The agent checks the signatures

    Before running, it verifies the signature of the flow, the launch, the catalog and the data row. If one fails, the job is blocked and says which.

  • Production, closed by default

    The agent only runs in production the flows its own configuration allows, whatever the platform asks.

  • AI off by default

    Your organization decides whether to turn it on and with what quotas. The model receives references and codes, never values or secrets.

The gate, inside your pipeline

GitHub Actions, Azure Pipelines or Jenkins launch the plan of a frozen release for a build, wait for the result and read the gate.

  • With the “Report” policy, the gate is reported and the pipeline continues.
  • With “Block”, the pipeline fails if the gate does not pass.
  • A pipeline token cannot write cases, approve or publish.
The TestEdge step in a pipeline
  1. LaunchPOST /api/v1/ci/scopes/{scope}/runs202 runId
  2. WaitGET /api/v1/ci/plan-runs/{runId}status: FINISHED
  3. DecideGET /api/v1/ci/scopes/{scope}/gateresult: FAIL, failPipeline: true

With the “Block” policy, the pipeline fails here and the log says which rule did not pass.

Release gate: does not pass

  • PassesEvery critical requirement met
  • FailsEvery mandatory point with a conclusive result
  • FailsNo mandatory point failed

Evaluating is not deciding: releasing is still a person's decision.

Frequently asked questions

Do I need to open ports or create a tunnel into my network?

No. The agent starts every connection to the platform over HTTPS. The platform never calls into your network.

What can be tested?

Web portals in the browser and REST APIs, including web tests that call an API in the middle of the journey.

Where is the agent installed?

On a machine in your network running Linux, Windows or macOS that can reach the environments you want to test.

What is a concurrent run?

One test running at a given moment. You subscribe to concurrent runs, not to installed agents or people: install as many agents as you need.

How is a journey recorded?

With the recorder, which opens its own browser on the tester's workstation, in a test environment. The recording is reviewed right there before anything is sent, and arrives as a draft your team completes with its checks.

How do I subscribe?

With an annual subscription. Write to us and we will prepare a demo with a case from your operation.

Try TestEdge with one of your cases

We show you the whole path: from requirement to evidence, with an agent in a test environment.

What you send goes to the PhoenixCode team, who make TestEdge AI. Privacy

Try TestEdge with one of your cases

For example: online banking transfers and their API, in a test environment.

With the country code, for example +1 786 555 0100.