Replay QA: Autonomous QA Agent for AI-Built Apps

Replay QA: AI wrote the app — this finds what broke

Replay QA is an autonomous QA agent for web apps, built by Replay.io on its original core technology: a deterministic time-travel recorder for software execution. Give it a URL and it explores the app on its own, finds bugs, records complete sessions, and returns root-cause analysis with confidence scores and suggested fixes — so your coding agent gets a full recording of what happened, not a screenshot and a guess. The homepage states the thesis plainly: "AI wrote the app. Replay QA finds what broke."

Founded around 2020 by ex-Mozilla Firefox engineers (CEO Brian Hackett, a Stanford PhD with a decade on Firefox), and backed by roughly $18.7M from Andreessen Horowitz and others, the company pivoted from its developer time-travel debugger to autonomous QA as AI-generated apps flooded in — typically shipped by solo founders, small teams, and agencies with no test coverage at all.

Key Features

Autonomous exploration of any web app from just a URL — zero setup for one-time testing
Session recordings with runtime breakdowns and confidence-scored root-cause analysis
Automatic Playwright test generation from discovered flows
GitHub integration for continuous testing; a CI agent that watches PRs and diagnoses failures
Authenticated-flow support and localhost/private-app testing via reverse proxy
Bug-report webhooks into Linear, GitHub Issues, and Jira, plus a Replay MCP for coding agents

Ideal Use Case

A solo founder or small team shipping a Lovable/Bolt/Cursor-built app that works on the happy path and has zero tests: point Replay QA at the staging URL before launch, hand the recordings to your coding agent, and let the generated Playwright tests become your regression suite.

How Replay QA differentiates

Playwright and Cypress require you to write tests; Replay QA explores autonomously and emits Playwright tests as output. Against AI-QA rivals like QA Wolf and Momentic, its bug reports carry deterministic time-travel recordings of actual runtime execution — evidence a coding agent can act on, not just video. Production credibility: a16z-backed, built by Firefox veterans with Redux maintainer Mark Erikson on the team, and homepage endorsements from engineers at Vercel, Glide, and Pantheon plus React maintainers Dan Abramov and Sebastian Markbåge.

FAQ

How much does Replay QA cost? The free tier includes 25 credits/month with no credit card; paid plans start at $20/month for individuals and $200/month for teams.

Do I need to write any tests? No — the agent explores from a URL, and it can generate Playwright tests automatically from what it finds.

Can it test apps behind login or on localhost? Yes — authenticated flows are supported, and localhost/private apps work via a reverse proxy.

Is this the same Replay as the time-travel debugger? Same company and same recording technology, repositioned in 2025-2026 as an autonomous QA agent for AI-built apps.

tl;dr

Replay QA turns the strongest debugging tech in the industry — deterministic time-travel recording — into an autonomous QA agent: URL in, bugs out with root causes and Playwright tests, free to try at 25 credits/month.

Looking for more options? Browse the Developer Tools directory or read our best AI coding tools listicle. Replay QA is also tracked on Crunchbase.