Skip to content
01Why does the app need the microphone all the time?
It does not. The microphone runs only during a session you start yourself, and it stops when you stop the session or when the session time limit runs out. While it runs there is a notification you cannot miss.
02Does it record my conversations?
No. Audio lives in memory for a few seconds and is discarded. Your own words are turned into text on the phone only to be counted; that text is not kept, nothing is written to storage, and nothing is uploaded. What is saved is a counter — a category, a time, a confidence value and the number of the dictionary entry that matched.
03How does it know it is me speaking?
From the short voice sample you record once. Speech that does not match it closely enough is discarded. In a noisy room, from a pocket, or with several people talking at once, matching gets harder — so events the app is unsure about are marked as such and never affect your index.
04It counted something that was not a swear word.
Tap the event and mark it wrong. It stops counting immediately. If it happens often, lower the sensitivity in Settings or add the word to your own list.
05Can I see what someone else says?
No. Kyndsy analyses one person — the owner of the phone. There is no mode for anything else.
06How do I cancel the subscription?
Through your Apple or Google account, in the subscriptions section — the same place all your other subscriptions live. The store cancels it, and your payment details never reach us.
07How do I delete everything?
Settings → Your data. “Delete voice profile” removes the voice profile; “Delete all history” removes every event and diary entry. Both are immediate and cannot be undone. Deleting the app removes everything on the phone; what our server keeps for a limited time — AI replies, a notification token — and when it goes is listed in the privacy policy.
08It said my tone was sharp, but I was only calling someone in the other room.
Loudness alone cannot tell those two apart, so the app also looks at whether anyone else has spoken in the last minute, how long the burst was, whether the sound was held or clipped, and whether a rough word was in it. One loud moment brings advice only if a rough word was in it; otherwise only a pattern that repeats within half a minute does. If it still gets it wrong, mark the event wrong: each mark raises the bar, and after a few the app speaks up about your tone only at the clearest outbursts. The sensitivity setting is about matching your voice, not about tone.