What QA teams hire Relyv to do
QA engineers spend the majority of their day on three things: reproducing reported bugs, writing regression tests, and verifying fixes. Relyv collapses all three into a single workflow built on pixel-perfect session replay.
- Reproduce any reported bug from the exact session, exact device, exact moment — no more "works on my machine".
- Auto-generate a Playwright or Cypress spec from any captured flow in one click.
- Diff a staging session against a production session — DOM, network, console — line by line.
- File the bug into Linear, Jira, GitHub, GitLab, Asana, ClickUp, Notion or 7 more trackers, pre-filled with the AI repro, evidence, and a runnable test.
Why session replay beats manual repro
A manual repro is a hypothesis: "the user said they clicked X, so it must have happened at Y." A Relyv replay is the truth: the exact DOM the user saw, every network call, every console error, every rage click, played back as the actual live page. You stop guessing and start solving.
Full DOM, not a video
Other "session replay" products record screen video — a flat MP4 you can watch but not inspect. Relyv reconstructs the actual DOM. You can open your browser dev tools inside the recording, hover any element, see what state was in the page, replay the console errors, and inspect the network waterfall.
Stage vs Prod diff
Pick a session from staging and a session from production with the same flow. Relyv shows you a side-by-side DOM, network, and console diff. If the bug is a regression introduced in the last release, the delta lights up in seconds.
Cross-session intelligence
"147 sessions hit the same checkout error this week, $13K at risk, root cause at line 42." Relyv surfaces the recurring failures across thousands of visits so QA can prioritise the bug that actually moves revenue — not the loudest ticket.
From session to runnable test in one click
The "auto-generate test" button is what QA leads tell us is the single highest-leverage feature. Pick any session, click Generate test, choose Playwright or Cypress, and you get a runnable spec with locators, assertions, and waits derived from the actual user behaviour. Tests cover the locator-strategy edge cases manual writing usually misses (data-testid first, then role, then accessible name, then nth-of-type as last resort) and self-heal when selectors drift.
- Playwright or Cypress export, your choice.
- Locator strategy: data-testid → ARIA role → accessible name → text → CSS — never coordinate-based.
- Self-healing: when a selector breaks, the AI repairs it from the live DOM.
- Imports cleanly into your existing repo; no new test runner required.
Privacy + compliance the QA team can actually defend
PII is masked on-device before any bytes leave the browser. The SDK uses 12-category regex (email, phone, SSN, credit-card with Luhn validation, etc.) and the optional Relyv Chrome extension adds an on-device LLM (GLiNER on ONNX/WebGPU) for free-form PII like names embedded in support chat. EU + US data residency available. SOC 2 Type II in progress.
How it compares to traditional QA tooling
"Traditional QA tooling" today means a Jira board full of "can't reproduce" tickets, a manual Playwright suite that breaks every sprint, a Loom recording of the bug that doesn't help engineers debug, and a heatmap tool that captures the user's session but not the developer context. Relyv replaces all four.