How Fluent handles your data

This notice is public. You do not need an account to read it. Fluent runs no third-party analytics or advertising scripts.

How long each kind of data is kept

Data	What happens to it
Speaking audio recordings	Stored privately. Removed after 24 hours, or up to 30 days if you turn on extended playback. Removed sooner when you delete the recording, withdraw recording consent or delete all practice.
Speaking attempts, their transcripts and AI speaking grades	Saved with the speaking attempt they belong to. They expire with that attempt (24 hours by default) and are removed by scheduled cleanup, or sooner when you delete them.
Written responses, saved revisions, feedback, review notes, mistake practice, goals, schedules, home-check answers, mock answers and mock results	Kept in your private Fluent workspace until you delete them. Fluent sets no automatic expiry. You can export them first.
AI writing feedback	Shown in your browser view only. Fluent does not save the AI grade; your saved response stays unchanged.
Live conversation audio, what you say and captions	Fluent does not save them. Captions exist only on your screen and disappear when you leave the page. Fluent keeps a usage record (role, time used, number of turns) so the daily and monthly limits work; it contains none of your words.
Practice you share	Each share expires after 7 days and you can revoke it sooner.
Anonymous funnel counts (event name, tool name, page path, campaign labels)	Not linked to your account, so they cannot be found or deleted from here. Fluent sets no automatic expiry yet.
Hosting logs	The hosting platform keeps request logs, including one line per funnel count. Fluent does not set its own log retention.
Deletion receipts and access or usage accounting	Minimized records stay to prevent replay and incorrect charges. They contain no practice responses.

Who processes your data, and where

Supabase hosts sign-in, the Fluent database and private audio storage, in Canada (AWS ca-central-1).
Anthropic (Claude) and OpenAI receive data only when you ask for it: the text of a written response for AI writing feedback, and the transcript of a saved speaking recording for AI speaking feedback. Fluent uses Anthropic first and OpenAI as the backup, so either one can process a given request. Only OpenAI receives a saved speaking recording itself, for transcription, and microphone audio in voice practice, which Fluent does not save. In live conversation practice, OpenAI hears your microphone audio as you speak and Anthropic (or OpenAI as the backup) receives the text of what you say to write the partner's replies; Fluent saves neither. Fluent does not restrict either provider's processing to Canada, so these requests can cross the Canadian border and the information can be subject to the laws of the place where it is processed.
Stripe takes card payments. Fluent never sees or stores your card details.

Cookie and counts

Fluent sets one first-party cookie, fluent_utm, only when you arrive through a link with a utm_source label and your browser is not sending Global Privacy Control or Do Not Track. It holds up to three campaign labels (source, medium, campaign) for 30 days so sign-up and checkout keep the channel you came from. It is not used for advertising.
When you use a free tool, start a practice attempt, finish an attempt or start checkout, Fluent counts that event on its own server with the event name, the tool name, the page path and those campaign labels. It records no account ID, email, IP address, transcript or free text, and sends nothing when your browser sends Global Privacy Control or Do Not Track.

Your private CELPIP practice

Privacy & recordings

Choose what you record and who can see your practice. Your Navisa account stays available when you delete Language practice.

Checking your current account…

Loading your Fluent workspace…

How Fluent handles your data

This notice is public. You do not need an account to read it. Fluent runs no third-party analytics or advertising scripts.

How long each kind of data is kept

Data

What happens to it

Speaking audio recordings

Stored privately. Removed after 24 hours, or up to 30 days if you turn on extended playback. Removed sooner when you delete the recording, withdraw recording consent or delete all practice.

Speaking attempts, their transcripts and AI speaking grades

Saved with the speaking attempt they belong to. They expire with that attempt (24 hours by default) and are removed by scheduled cleanup, or sooner when you delete them.

Written responses, saved revisions, feedback, review notes, mistake practice, goals, schedules, home-check answers, mock answers and mock results

Kept in your private Fluent workspace until you delete them. Fluent sets no automatic expiry. You can export them first.

AI writing feedback

Shown in your browser view only. Fluent does not save the AI grade; your saved response stays unchanged.

Live conversation audio, what you say and captions

Fluent does not save them. Captions exist only on your screen and disappear when you leave the page. Fluent keeps a usage record (role, time used, number of turns) so the daily and monthly limits work; it contains none of your words.

Practice you share

Each share expires after 7 days and you can revoke it sooner.

Anonymous funnel counts (event name, tool name, page path, campaign labels)

Not linked to your account, so they cannot be found or deleted from here. Fluent sets no automatic expiry yet.

Hosting logs

The hosting platform keeps request logs, including one line per funnel count. Fluent does not set its own log retention.

Deletion receipts and access or usage accounting

Minimized records stay to prevent replay and incorrect charges. They contain no practice responses.

Who processes your data, and where

Supabase hosts sign-in, the Fluent database and private audio storage, in Canada (AWS ca-central-1).

Anthropic (Claude) and OpenAI receive data only when you ask for it: the text of a written response for AI writing feedback, and the transcript of a saved speaking recording for AI speaking feedback. Fluent uses Anthropic first and OpenAI as the backup, so either one can process a given request. Only OpenAI receives a saved speaking recording itself, for transcription, and microphone audio in voice practice, which Fluent does not save. In live conversation practice, OpenAI hears your microphone audio as you speak and Anthropic (or OpenAI as the backup) receives the text of what you say to write the partner's replies; Fluent saves neither. Fluent does not restrict either provider's processing to Canada, so these requests can cross the Canadian border and the information can be subject to the laws of the place where it is processed.

Stripe takes card payments. Fluent never sees or stores your card details.

Cookie and counts

Fluent sets one first-party cookie, fluent_utm, only when you arrive through a link with a utm_source label and your browser is not sending Global Privacy Control or Do Not Track. It holds up to three campaign labels (source, medium, campaign) for 30 days so sign-up and checkout keep the channel you came from. It is not used for advertising.

When you use a free tool, start a practice attempt, finish an attempt or start checkout, Fluent counts that event on its own server with the event name, the tool name, the page path and those campaign labels. It records no account ID, email, IP address, transcript or free text, and sends nothing when your browser sends Global Privacy Control or Do Not Track.