TestivAI Logo

Our Workflow

The Test, Analyze & Approval Loop

TestivAI is designed to fit seamlessly into your existing CI/CD process. Here is the end-to-end journey of a single visual test.

The Diff & Approval Process

This is the core loop you'll follow for every visual change.

Visual Workflow
TestivAI Smart Match Process Diagram

The complete flow from baseline creation to approval

  1. 1

    Run 1: The Baseline is Created

    The first time you run your test, the @testivai/witness package sees that no "golden" baseline exists. It takes a snapshot, captures the DOM, and sends it to the TestivAI API to be stored as the official baseline for this test.

  2. 2

    Run 2: A Change is Detected

    A developer pushes new code. The test runs again. This time, @testivai/witness takes a new snapshot and compares it to the baseline. It finds a difference.

    This is the critical part: Instead of sending the full (and sensitive) screenshot, it only sends the small diff image and the isolated DOM snippet of the changed element to our API. Your CI build is immediately unblocked and finishes fast.

  3. 3

    Analysis: The AI Judge Investigates

    In the background, our AI Worker picks up the job. It runs the full **Smarter Match** analysis on the diff and the DOM context. It determines if the change is a real bug (e.g., "Button text is overlapping") or a dynamic content change (e.g., "News headline updated") and writes its verdict to the database.

  4. 4

    Review: See Results in the Dashboard

    You open the TestivAI dashboard and see the new test run, flagged as AI_BUG. You see the diff image, the isolated DOM, and the AI's plain-English root cause analysis. There are no pixels to guess at—just clear, actionable results.

  5. 5

    Action: Approve the New Baseline

    You see the change was intentional (e.g., a button color was updated). You click the "Approve" button in the dashboard. This sends a secure request to our API, marking the snapshot from this test run as the new "golden" baseline.

  6. 6

    Run 3: The Test Passes

    The next time your CI pipeline runs, @testivai/witness fetches the newly-approved baseline. It performs its comparison, finds no differences, and the test passes instantly. The loop is complete.

Stop Chasing Pixels.
Start Shipping with Confidence.

Join the waitlist for early access.