autony

Real-phone testing for apps your agent builds.

Describe the core flow in one sentence. A tester uses your release APK on real Samsung phones, runs the same path again with the network off, fonts at 130% and the phone in the locale you choose (English, Korean or Japanese), and returns a report your coding agent can act on. No scripts, no SDK, no device slots.

See a real report 
claude mcp add --transport http autony https://autony.connect-y.com/mcp \
  --header "Authorization: Bearer autony_sk_…"
> "upload app-release.apk to autony and test the core flow"

No coding agent? Upload the APK in your browser — same report. · your first Quick Scan is free · prices below

A goal, not a script

“Create a note and save it.” The tester reads the screen and finds the way, step by step, until it's done. One named flow, one phone per run, assigned automatically.

Run again under real conditions

The same path, again, with the network off, fonts at 130% and the phone in the locale you choose. On our Galaxy phones (Android 13–15), not an emulator.

A report built for agents

Markdown with evidence links for Claude Code or Cursor; HTML with screenshots for you. Your app's text can't steer the tester.

What a report looks like

A verdict, defects ranked by how much they hurt the user, each with before/after screenshots, the log excerpt and the steps to reproduce — and the exact path the tester took, so “no defects” still shows what was covered.

Open the real, unedited sample →

example defect card · from the regression app our lab runs daily (a known offline bug)P1 · blocks the flow
Screenshot: the save screen right before the offline tap
Offline handlingOnly a progress indicator with no notice for 14.9 s while offline.Galaxy A15 · the screen right before the offline tap at step 5 · before/after pair and log excerpt attached

How it compares

CriterionDevice clouds (Firebase Test Lab, BrowserStack)AI test authoring (e.g. Maestro, QA Wolf)autony
You writeAppium / Espresso scriptsTest steps in plain EnglishOne sentence
Who runs itYour scripts, their phonesTheir AI, often their phonesOur tester, our phones, offline / fonts / locale re-runs
OutputPass/fail, videoPass/fail, screenshotsDefects with severity, repro steps and evidence, Markdown for your agent
PricingPer device slot or minutePer seat or creditsPer scan, prepaid credits: Quick Scan from $20, Launch Scan from $100
Coming from Firebase Test Lab Robo?Robo shuts down on Sep 30, 2027. Robo crawled randomly across a device matrix; autony follows one flow you name, on one phone per run, and re-runs it under conditions Robo never covered. Upload the same APK, no test package needed.

Pay per scan with credits. No seats, no slots.

Quick Scan · 1 creditOne mid-range phone, about 5 minutes once it starts: your core flow, offline and 130% font re-runs, crashes and freezes, accessibility. For any build you want checked.
Launch Scan · 5 credits Coming soonThree phones (budget, mid-range, flagship · Android 12–15), up to 3 languages, and one re-scan of the same flow within 14 days to confirm your fixes. For a launch or a major update.
Free$0Your first Quick Scan
3 credits$79$26 per credit · 3 Quick Scans
10 credits$229$23 per credit · 2 Launch Scans
30 credits$599$20 per credit · 6 Launch Scans

Credits never expire. A scan that reaches a phone is charged, even with no defects. If the install fails, our lab can't complete the scan, or the app stops at a sign-in screen (during the beta), its credits are returned. Unused packs are fully refundable within 14 days of purchase; credits already used are not refunded.

Prices in USD; payments processed by Paddle. Beta: we approve new accounts within 1 business day, then your free credit is ready.