Recording and self-healing · updated 2026-10-05

Test recorder and self-healing automation

Record a flow in a real browser, replay it as a regression test, export Playwright, Selenium or Robot Framework code, and approve every heal in a review queue.

For: QA engineers and SDETs who maintain regression suites, and analysts who need to create tests without writing code.

Start free Book a demo on your own flow

Capabilities

Real browsers, five engines

Chromium, Chrome, Firefox, WebKit and Edge. The recording runs in the browser your users use, not an emulation of it.

Seven structured files per session

Session metadata, element details, network data, performance data, security data, UI elements and a replayable step list, with stable cross-reference IDs between them.

Self-healing with a review queue

More than fifteen fallback locator strategies propose a repair; a person approves or rejects it; the element repository learns from approvals.

Exports to code you own

Playwright, Selenium or Robot Framework. Migration Setup imports existing Selenium, Cypress and Playwright assets with a per-item report.

Folders, schedules and a grid

Group runs in folders, run them on a schedule or from a trigger API, and execute across browsers in parallel on the execution grid.

Smart test selection

Run only the scenarios a change or a requirement touches, within a time budget, with a written rationale per pick.

Point the recorder at a URL you are allowed to testNothing is installed on your machine; the browser runs on the platform.
Click through the journeyEach step records the element, its locators and the network calls it caused.
Add assertionsText, visibility, URL, response status: whatever the step must prove.
ReplayThe recording runs as a regression test immediately, step by step, with screenshots.
Schedule, gate or exportRun it on a schedule, fail a pipeline on it, or export it as code and keep it in your repository.

What you receive, and what we do not do

Evidence ledger

Per session: Seven structured files with stable IDs: metadata, elements, network, performance, security, UI elements, replayable steps.
Per step: The element targeted, every locator available for it, the network calls it caused, the assertion result, a screenshot.
Per heal: The proposal, its reasoning, who approved it and when; nothing applied without a person.
Per run: Pass or fail per step, duration, browser, and the export of the same flow as code.

Limits we state

Heals are never auto-approved; a fully unattended suite still needs someone to review the queue.
A recording captures what the browser sees, including inputs. Record against test accounts and synthetic data; recognised credentials are redacted at capture time.
Export to Cypress is not available; import from Cypress is.
We publish no maintenance-saving percentage. The product shows you the measurement on your own runs.

Plans: Every plan, including Free (100 recordings a month). Solo Pro 1,000; Team 10,000, pooled across the workspace; Enterprise unlimited. See the metering table.

Go deeper

Questions buyers ask

Do I need to install anything?

No. The browser runs on the platform; you point it at a URL you are permitted to test and click through the flow from your own browser.

What happens when the UI changes?

The step that can no longer find its element gets a proposed repair from the fallback locator strategies. It waits in heal review until someone approves or rejects it; the test is not changed silently.

Is this Selenium or Playwright under the hood?

Execution is Playwright-based with multi-browser support, and recordings export as Playwright, Selenium or Robot Framework code, so you choose the format your team maintains.

Try it on your own application

The Free plan has no time limit and asks for no card. Or bring one flow from your product and we record it live.