FeaturesHow it WorksPricingDocsFAQ
Zero-code test automation

Stop Writing
Brittle Test Scripts.

Record your manual flows once. Let Joona Mercury's AI execute them automatically — no code, no semantic tags, no developer bottlenecks.

No credit card required No data-testid required Bring your own AI key
mercury · checkout-flow.session
Recorded once
01  click .checkout-btn
02  type #shipping-name
03  select .payment-method
04  click "Place Order"
Replayed by AI
DOM changed: .checkout-btn → not found
Matched by role + label: button "Checkout"
Generated test input for #shipping-name
All 4 steps resolved against current DOM
✓ Run passed · 1.8s
Why teams switch

Why QA Teams Switch to Mercury

Every benefit here removes a specific, recurring source of pain in test maintenance.

🚀

Zero Setup on Target App

No need to beg developers to add data-testid or change the DOM. Just record and run.

🧠

AI-Driven Execution

Our AI understands DOM context. If an ID changes, the AI finds the element anyway.

🔄

Record & Replay

Capture complex flows — login, checkout, multi-step forms — in minutes. Click "Run" infinitely.

🧪

Smart Data Generation

AI intelligently generates test input data for your forms. No more hardcoding dummy data.

💸

Predictable Costs

Flat-rate DOM-based execution. No hidden vision API costs. Know exactly what you'll pay.

🛠

Multi-Model Support

Bring your own API keys. Works with OpenAI, Claude, DeepSeek, and any OpenAI-compatible API.

The automation loop

From Manual to Automated in 3 Steps

01

Record Your Flow

Use our lightweight browser extension to click, type, and navigate as you normally do.

02

AI Learns the Context

We capture semantic DOM snapshots, parent HTML, and bounding boxes — not just pixel coordinates.

03

Run & Relax

Hit "Run" on the dashboard. Mercury executes the flow, generates data, and reports detailed results.

Proof of depth

Built for Modern QA Workflows

Every run feeds a live dashboard — so you see test health at a glance, not buried in CI logs.

mercury.app · dashboard · checkout-project
Total Flows
24
Total Runs
1,842
Success Rate
96.4%
Avg Duration
2.3s
Hours Saved
118h
vs. manual testing
Execution Trend · Last 7 Days
Mon
Tue
Wed
Thu
Fri
Sat
Sun
SuccessFailed
Flow Status
Healthy19
Failing3
Never Run2
Top Failed Steps
click · .promo-code-apply7 fails
type · #shipping-postal-code4 fails
select · .payment-method3 fails

Chrome Extension Recorder

MV3 compliant, lightweight, runs alongside your normal browsing.

Semantic DOM Snapshotting

Resilient to layout changes — captures meaning, not just position.

AI Selector Resolution

Finds elements even if IDs or classes change between deploys.

Dynamic Test Data Generation

AI fills forms intelligently, matching field type and context.

Detailed Run Logs & AI Reasoning

See exactly why a step passed or failed, step by step.

Multi-Provider AI Configuration

BYOK — bring your own key for OpenAI, Claude, DeepSeek, and more.

Flow Management Dashboard

Organize, edit, and schedule flows across your whole test suite.

Pricing

Simple, Predictable Pricing

Start free. Scale when you're ready.

Free
$0

For individual QAs getting started.

  • 5 flows
  • 50 runs / month
  • Core AI features
  • Community support
Most popular
Pro
$49/mo

For growing QA teams.

  • Unlimited flows
  • 500 runs / month
  • Priority support
  • All AI providers (BYOK)
Enterprise
Custom

For orgs with compliance needs.

  • Custom AI providers
  • SSO
  • Dedicated support
  • SLA
Social proof

Loved by QA Engineers

I used to spend hours fixing broken Selenium selectors every sprint. Mercury just works, even when the dev team changes the UI.

RA
Raka A.
SDET, fintech startup

We tripled our regression coverage without adding a single test-only attribute to our React app. That alone justified the switch.

SM
Sarah M.
QA Lead, B2B SaaS

The reasoning logs are what sold our team. When a step fails, we see exactly why instead of guessing at a stack trace.

DP
Dimas P.
QA Engineer, e-commerce
Technical FAQ

Questions, Answered

Yes — your account is what stores recorded flows, runs them on schedule, and connects your AI provider key. Sign up free first, then the extension prompts you to install in under a minute.

Yes. Mercury watches the rendered DOM, not the server response, so client-side routing and dynamic re-renders are handled the same way a real user would experience them.

The AI re-resolves each step against the current DOM using semantic context — role, label, position, surrounding text — rather than a fixed selector, so most layout and styling changes don't break your flow.

No. Mercury runs entirely from the browser extension and our dashboard. Your application code and infrastructure stay untouched.

Snapshots include parent context and bounding boxes alongside the semantic tree, which lets the resolver pierce shadow boundaries and disambiguate elements that share similar markup.

Yes. Mercury is BYOK by default — connect OpenAI, Claude, DeepSeek, or any OpenAI-compatible endpoint, and your DOM data is sent only to the provider you choose.

The run stops at that step and the dashboard shows the AI's full reasoning trace — what it tried, what it matched, and why it couldn't proceed — so you can fix the real cause, not guess at it.

Ready when you are

Reclaim Your Time.
Stop Fighting Selectors.

Join the QA engineers who automated their repetitive flows in minutes, not weeks.

Free forever for 5 flows. No credit card needed.