What this guide covers
OKYA Watch works with pixels from a screen, window or tab that you choose to share. It can turn visible text into a record, read a number or watch a visual condition. Start with a harmless sample rather than a private account screen. That makes the first result easy to check and avoids exposing unrelated information while learning the controls.
Set it up
Open the monitoring workspace and add one source. Choose the intended tab or window in the browser picker, then confirm that its preview is live. Create a task for a small region, select text recognition and the language used in that region. Enable a record or in-page notification action, save the task and start recognition on that source.
Check the result
Use a known phrase with a clear font. Change the phrase once and compare the recognized result with what is visible in the source. A successful OCR result does not automatically mean a notification rule matched: the detector, condition and action are separate choices. Check the task state and the execution result in history rather than relying on the presence of a preview alone.
Before you rely on it
Source configurations and active connections are different. You can save up to eight configurations but connect at most four sources simultaneously. A saved source name does not preserve a browser sharing permission after refresh. Reconnect it when required, and verify that tasks still point to the intended source before trusting a new session.
Troubleshooting
If the preview is missing, solve sharing permissions first. If the preview is clear but text is wrong, adjust the region or OCR language. If recognition is correct but there is no alert, inspect the condition and selected action. Test one layer at a time; changing every setting together makes an intermittent problem much harder to identify.
Product reference
The local demonstration uses generated pixels and labels its records as samples. It is useful for learning the interface, not proof that your application will render correctly in the background. Complete a second test with the actual source, including one condition that should match and another that should not. Stop sharing when the session is over.
No installation or sign-in is required. Tasks and recognition history stay in this browser. You still need to authorize sharing a window, tab or screen before recognition starts.
Further reading
- Screen sharing and browser permissions
- Local OCR models and recognition languages
- Keep, review and export local recognition history