AI-assisted test automation
Ship on Friday.
SmartestQA tests your web, mobile and desktop applications without code, and tells you what passed, what failed and why. Record a test, build it from reusable actions, or let Claude write it.
No code to write. Playwright, Selenium and Appium underneath.
The film / one minute
Tonight, you sleep.
Release night, the way it too often goes: a test that fails for no reason, a red pipeline, and the question nobody wants to hear. Then the way it ends with SmartestQA.
One question. One answer.
READY
Every layer passed. Nothing hidden. Ship it.
Illustration. The release and the numbers are examples. Scroll to play it.
What it is
One platform. Every layer of the test.
SmartestQA is an AI-assisted, no-code test automation platform made by Selenity Technology. It clicks through a web page, taps a phone, calls an API, queries a database, reads an email and runs a load test. Pipelines run the scenarios on a schedule and keep a report of every run.
- Applications
- Web, Android, iOS, Windows and macOS desktop
- Engines
- Playwright or Selenium for web, Appium for mobile and desktop
- Runs on
- The Smartest Server, your own agent, or a device farm
- AI assistants
- Claude, Claude Code, Cursor and any other MCP client
MCP
Ask Claude. Approve the diff.
Connect Claude, Claude Code or Cursor to your test project. The assistant sees where the project stands, writes the tests, runs them and explains a red run. It changes nothing until you approve.
Add the checkout requirements to project Web Shop.
Proposal · nothing written yet
Proposed for project Web Shop: + MAIN Checkout module: Orders + SUB Pay with a saved card under: Checkout + SUB Card is declined under: Checkout Impact: 0 existing records change. Apply this?
Why it exists
A green run can lie in three ways.
None of them shows up on a dashboard. All of them show up in production. SmartestQA is built to catch each one.
01 / The empty green
A test that checks nothing passes forever.
It clicks through checkout and never checks a result. It stays green on the day checkout breaks.
Test Health counts a requirement as covered only when a linked test checks a result and passed its last run. A test without a proof step gets a trust warning.
02 / The heal that hides a bug
Self-healing has to know when not to heal.
A tool that always finds something similar clicks Cancel when Save has disappeared, and the run stays green.
Auto-heal checks the page first. On a broken page it refuses and the step fails. It proves that the repaired step still did its job, and every repair waits for your review.
03 / The maintenance tax
Fix a control once. Every test gets the fix.
In a hand-written suite a renamed button breaks every script that touches it, one locator at a time.
A recording becomes reusable actions with several locators each. One fix repairs every test case that uses the control.
How it works
Seven modules. One order of work.
Each one hands its result to the next.
Discover
SmartestQA learns your application: its pages, the controls on each page, how the pages connect and the business rules nobody can see on screen.
MoreAnalyze
You write down what the application must do, by hand or from a document with AI, and link each rule to the test cases that prove it.
MoreAutomate
You build the tests: record a flow or describe it, feed it test data and try it at once.
MoreSchedule
Tests become runs. A pipeline runs test cases on demand, on a timer, from CI/CD or from a trigger.
MoreReport
Every pipeline run lands here with its steps, screenshots, video, logs and an AI verdict on each failure.
MoreMaintain
What your runs found waits here for a decision: repaired locators, AI steps to turn into actions, network captures to turn into API tests.
MoreInSight
The dashboard. Execution shows how your runs went. Test Health shows how much of your application your tests really prove.
More
Features
Five things teams ask about first.
- MCP
Write and run tests from Claude
Claude, Claude Code and Cursor plan, write and run tests on your real project. Nothing changes until you approve it.
MCP and AI assistants - Mobile
Real devices in your browser
Real Android and iOS devices, live in your browser and shared with your pipelines.
Device Farm and mobile testing - Record
Record once. Reuse everywhere.
Do the flow once. It becomes a reusable test that repairs itself when your application changes.
Record and replay - API
Your UI tests already know your API
The calls your browser tests already make become API tests, without writing them.
Network Calls to API tests - Load
A k6 or JMeter test from the API steps you have
Describe the load in a sentence. Run it, and see whether you got slower.
Load testing with k6 and JMeter - Use cases
One question, answered start to finish
Prove that every requirement is tested. Build a regression suite and run it every night. Each use case gives the short answer and the steps.
Use cases
One scenario. Every layer.
Chain browser, mobile and desktop actions with API, database, mail, Kafka, code and load-test steps in the same test case.
- Web
- Android
- iOS
- Windows
- macOS
- API
- Database
- Kafka
- Python
- JavaScript
- AI steps
- JMeter
- k6
Runs where your app is.
A public web app runs on the Smartest Server with nothing to install. An app inside your network runs on your own agent, and only results go back to SmartestQA. Phones run on your agent or on a device farm.
- Smartest Server
- Your agent
- Device Farm
- AWS Device Farm
- BrowserStack
Fits the tools you have.
Sync tasks and test cases, publish reports, alert the team and start pipelines from your CI.
- Jira
- Azure DevOps
- Xray
- Zephyr Scale
- Confluence
- Slack
- Microsoft Teams
- Webhooks
- CI/CD
- Datadog
- Dynatrace
Compare
What teams compare it with.
Each approach is a fair choice in some situations. This is where SmartestQA differs.
| Approach | Strong when | Where SmartestQA is different |
|---|---|---|
| Hand-written Selenium or Playwright code | You have automation engineers and want full control in code. | No code to write or maintain. Testers and analysts build tests; the engine still runs Playwright or Selenium underneath. |
| Record-and-playback tools | Quick one-off scripts. | Recordings become reusable actions with several locators each, so one fix repairs every test that uses the control. |
| Separate tools per layer (UI, API, database, load) | Each team owns its own layer. | Web, mobile, desktop, API, database, mail and load-test steps can be chained in one scenario. |
| Manual regression testing | Rarely changing apps or exploratory work. | Repeatable runs on a schedule, with video and screenshots for every step and requirement traceability. |
FAQ
Questions people ask first.
What is SmartestQA?
SmartestQA is an AI-assisted, no-code test automation platform made by Selenity Technology. It tests web, Android, iOS and desktop applications, and one scenario can also hold API, database, mail and load-test steps.
Do testers need to write code?
No. Testers and analysts record flows and build test cases from reusable actions. The engine runs Playwright or Selenium underneath. When you want code, a step can hold Python or JavaScript.
Does a test need AI to run?
No. AI is optional per project. A recorded step runs from its saved locators. AI names controls, writes requirements and steps, explains failures and repairs locators when you switch it on.
Which AI assistants can work with SmartestQA?
Claude on claude.ai through the connector, Claude Desktop through the extension, Claude Code and Cursor through a one-line install, and any other MCP client that can start a local server. See MCP and AI assistants.
Where do my tests run, and does my data leave our network?
With Run on Server, the browser runs on SmartestQA's servers and opens your app from there. With an agent, the browser runs on your machine inside your network, and only results (logs, screenshots, video) go back to SmartestQA. Use agents when the app must not be exposed to the internet.
Which browsers are supported?
You use SmartestQA and record in Chrome or Edge. Tests run in Chrome.
Which languages does the interface support?
English, Turkish, German and Spanish. The documentation is in English and Turkish.
How much does SmartestQA cost?
Plans are customer specific. Contact us and we will set one up for your team.
Blog
How to do it, step by step.
- How to write and run tests from Claude with MCPConnect Claude, Cursor or another MCP client to your test platform, let it plan, write and run tests, and approve every change first.
- How do self-healing tests work, and can you trust them?Self-healing tests repair broken locators at run time. You can trust them only when they refuse on a broken page and show you every repair.
- How to test a mobile app on real devices from the browserA device farm puts real phones in your browser and lets automated tests run on the same devices. How it works, shown with SmartestQA Device Farm.

