Define success in terms of the learning activity
For a chat exercise, success includes receiving the intended response, not merely seeing an input field. For a saved card, it includes confirming the save action rather than only reading a suggested word. Identify the result you need before deciding that an app is working normally on a weak connection.
This distinction is useful beyond installation. A browser tab can display a page while a live request fails, and an app-style window can do the same. The window shape does not tell you whether a particular learning action has reached the service. Watch the actual response and any visible status the interface provides.
Separate three commonly confused states
A familiar screen, an attempted submission, and a completed response are not interchangeable. Keeping them separate helps you avoid losing a draft or repeating an action unnecessarily. Start by describing what you can directly verify rather than assigning a broad label such as “everything is offline.”
| What you observe | What it establishes |
|---|---|
| The Unihongo entry opens | You reached a window or page. |
| You pressed send | You attempted an action; confirm its result. |
| The expected response appears | That particular request completed sufficiently to show its result. |
Preserve the learner’s work before changing the page
If you have written an important response, keep a safe copy before reloading or closing a window. Do not assume a draft exists on another device or that an interrupted request was saved. Avoid clearing browser data as the first troubleshooting move, because that can affect local sessions and settings while leaving the underlying connection problem unresolved.
Restart with a small connected check
After connectivity improves, check the account and try a modest activity whose result is easy to verify. Then inspect the previous task to see what actually completed. If the state is unclear, describe the last confirmed response before asking Mocchi for a follow-up. This is more reliable than reconstructing progress from the number of times you pressed a button.
Choose an offline fallback without inventing app capabilities
You can use your own already-available notes to review a distinction or rehearse a sentence while waiting for a connection. Keep that separate from a claim that Unihongo’s AI is operating offline. When service access returns, bring the sentence back for feedback. The learning can continue flexibly while the limits of the connected tool remain clear.