Your curiosity deserves clarity.

About Flowst Air

Flowst Air is an independent learning environment. Misu proposes plans for your approval; Amina guides spoken practice. Amina is an AI learning partner, not a human tutor or accredited examiner. Upload a resource and practise it aloud in a guided conversation. Check important answers against the cited source.

How learning guidance works

Before uploading, choose your purpose, focused or broad coverage, and available time. Add a short brief if you need a particular focus. A planning service uses these choices and your material to propose source-backed objectives and estimated effort. Review the plan before starting; the estimates do not promise a fixed finish time.

When you start a live call, Amina prepares an opening from your active objective, then responds aloud as you talk. You do not need to review a recording or press Send. Suggestions draw on saved evidence; completing an activity is not proof of mastery.

Your material and data

Your uploaded document or confirmed source snapshot is stored in cloud object storage. Extracted passages, session choices and any short brief, study plans, finalized transcripts, feedback, and learning activity records are stored in a cloud database. Study operations check the signed-in owner before accessing records. Authorized service operators may access data for operation and support.

AWS stores your study records and provides account and backend services. Source passages, your planning choices and brief, and conversation text are sent to Groq to prepare learning plans and replies. ElevenLabs processes live voice calls and any separate recorded exercises. This experience does not run entirely on your device.

Deleting a study chat removes its original document and associated study records from application storage. An active-session marker remains until its agreed objectives are completed or you explicitly abandon the plan. Deleting an unfinished document does not unlock a replacement by itself; Replace study document asks you to confirm abandonment, leaves any saved document read-only, and does not mark the objectives complete. No automatic retention period or deletion of provider-side records is promised by this interface.

Source links and video transcripts

Public repositories and readable web pages become attributed text snapshots. Review the included passages before creating a plan. Repository files are pinned to a commit; source content cannot give the agents new instructions.

Where video transcription is configured, choosing Create transcript sends the public YouTube or TikTok link to ElevenLabs. This uses a separate source allowance, not your voice-practice time. AI transcripts can contain errors. Only speech is included; diagrams, demonstrations and on-screen text are not inspected. Pasted or uploaded transcripts are labelled as learner-supplied.

Draft access expires after 30 minutes; storage cleanup may happen later. Cancelling a draft prevents its use but may not stop provider work already submitted. Deleting a study record does not delete the original public source or independent provider records.

Microphone and transcripts

Your microphone starts after you choose Start session or Start live call and grant permission. During the call, speak naturally and hear Amina reply aloud. Mute silences your input. End call, leaving the session, and sign-out stop microphone capture and playback. You can interrupt Amina while she speaks.

Each live pilot call lasts up to 60 seconds. Voice practice has no cumulative study or daily voice quota. Usage is still recorded; a failed or uncertain provider request can be included in that record.

Available live caption previews may change. Learner responses and generated replies appear as saved only after application storage confirms them. Transcripts can contain speech recognition errors. A saved generated reply does not prove you heard it completely, especially if you interrupted. Live call responses do not automatically advance assessed activities. Your microphone recording is not saved as an application audio file. Prepared Amina replies are saved privately for Replay and deleted with the study conversation. Replay of prepared audio does not send another synthesis request.

Accessibility

Controls support keyboard navigation and visible focus. Saved replies are readable in the conversation, and animation respects your device’s reduced-motion preference. Speech input requires microphone support and permission; typed input is not available. Available live captions remain separate from confirmed saved records.

Report a problem

If a request fails, use its retry control. For microphone problems, check this site’s permissions in your browser. For pilot support, contact the person who provided your access; include what happened without sharing private documents or recordings.