Privacy Policy

Familiai records conversations between two people, turns them into text, and helps both of them look back at what was actually said. That is an unusually sensitive thing to hand an app, so this page tries to be specific rather than reassuring.

The three things most worth knowing:

We store the recordings. The audio of conversations you record is uploaded to our servers and kept there. This is the most sensitive thing we hold, and we would rather say so first than bury it.
Google transcribes the recording, and it goes nowhere else. Your phone used to turn the recording into text, but it was not accurate enough to be worth trusting. That job now runs on Google Cloud, which is already where your recordings are stored — so no new company receives them. Google does not keep or learn from the audio: that logging is something a customer has to switch on, and we have not. The resulting text is stored on our server, and excerpts go to Anthropic to produce the analysis the app exists for.
You see the transcript after it is made, not before it is sent. That order changed when transcription moved off your phone, and it is the one thing here that got worse. If you do not want a conversation kept, delete it — the recording, the text and everything built from it all go.
Both people see everything. There is no private view. An archive shared by two people shows both of them the same thing, always.

What we collect

What we do not do

Consent, and recording other people

The app will not record until each person has heard what it does and agreed aloud, in their own voice, during setup. That recorded agreement is the consent record.

Data	Where it goes	Why
Conversation audio	Our server (Google Cloud, United States), encrypted at rest	Transcribed by Google Cloud, then kept so the text can be remade later as speech recognition improves. Your phone no longer transcribes — it was not accurate enough.
Voiceprint (a mathematical summary of your voice)	Stays on your device. Never uploaded.	Lets your phone recognise which parts of a conversation are yours.
Conversation transcripts	Our server (Google Cloud, United States)	The archive you search and ask questions of.
Excerpts of transcripts	Anthropic (Claude API)	Producing summaries, finding commitments, answering questions.
Email address and account identifier	Our server, from Google or Apple at sign-in	Identifying your account. Nothing else.
Your name, as you type it during setup	Our server	Labelling who said what.

Recording conversations is regulated differently in different places. In some jurisdictions every participant must consent; in others one is enough. You are responsible for knowing which applies to you. The app is built to require both, which is the stricter standard, but it cannot know where you are or who else is in the room.

Sharing between two people

An archive is shared by two accounts. Each person agrees to that explicitly when they join.

Either person can withdraw that agreement at any time. When they do, the archive closes for both of you after seven days. Nobody keeps a copy the other one does not have. During those seven days you can both still read and export everything, and whoever withdrew can change their mind.

Taking a copy

Either of you can download everything, at any time. Settings › Privacy › Download a copy of everything gives you one file containing every conversation, every transcript, everything the app worked out from them, and links to the recordings themselves.

Both of you get exactly the same file. That is deliberate, and it is the same rule as everywhere else here: this archive never shows one of you something the other cannot see.

Why this matters more than it sounds: this is a shared record, and either of you can delete it. Being able to keep your own copy first means a disagreement about the archive does not have to mean losing it. The links to the recordings expire after a while, so download them when you download the file rather than later.

Deleting your data

Your voiceprint is on your device, so deleting the app removes that. The recordings are on our servers and are removed by the deletions above.

Where your data lives, and for how long

Recordings and transcripts are stored in Google Cloud in the United States, encrypted at rest and in transit. Encrypted backups are retained for 14 days. We keep your archive for as long as your account exists; we do not have a separate retention schedule that deletes it on your behalf.

They were held in India until 16 October 2026. They moved because the speech recognition this app depends on is only fully supported in Google's US regions, and running it on a preview service was not a trade worth making for a record people rely on being accurate.

What changed, and when. Until 15 October 2026 this app kept audio on the phone, transcribed it there, and uploaded only the text. Two things changed on that date, and both are material:

If you agreed to an earlier version of this policy, please decide again whether you are comfortable. Withdrawing consent closes the archive for both of you, as described above.

Who else processes it

Children

Familiai is not intended for anyone under 18, and we do not knowingly collect data from children.

Changes

If this policy changes in a way that affects how your data is handled, we will say so in the app before the change takes effect.

Processor	What they receive
Google Cloud Platform	Hosting and storage of recordings and transcripts, and transcribing the audio into text. Google does not log or train on it — that is an option a customer must switch on, and we have not.
Anthropic	Transcript excerpts, for analysis. Not retained for training.
Google or Apple	Sign-in only. They are told you signed in; they are told nothing about your conversations.