Privacy
See where your interview data is processed, what stays on your PC, and the choices you control.
Where interview content is stored
Saved session transcripts and answers are stored on your PC, encrypted for your Windows account. During normal cloud inference, requests and responses pass through CueFlect's gateway in memory; CueFlect does not persist that interview content in server logs or stores. The inference provider's retention is separate. Content you choose to submit in a support ticket or attachment is stored with that ticket and is subject to support retention.
On the Full edition in Local mode, your interview content stays on your device.
Where your data goes, by mode
Scroll across to compare all three modes.
| Data or process | Local | Mixed | Cloud |
|---|---|---|---|
| Audio capture | Device | Device | Device |
| Normal/default transcription | Device | Device | Device by default, on either installation; CueFlect gateway only after explicit route selection |
| Text reasoning | Device | CueFlect gateway and cloud AI provider | CueFlect gateway and cloud AI provider |
| Screenshot pixels | Device only | Device only | Cloud allowed, under the accepted terms of use |
| Session history | Encrypted device storage | Encrypted device storage | Encrypted device storage |
What the cloud is allowed to do, and what you still choose
The app permits cloud processing under CueFlect's terms of use, which it presents before your first practice session and again whenever they materially change. They set out what is transmitted while a session is live, what is stored and what is not, and who is responsible for the other voice in the room. Sending the question to the cloud is not an optional extra on a plan that answers in the cloud — it is the service — so the terms disclose it rather than asking you to switch it on. Accepting those terms does not supply another participant's consent or establish a lawful basis for processing their information.
What you do choose is everything the product works without. Nothing from your profile reaches the cloud without an explicit approval bound to the exact text you approved. Editing approved text withholds it and tells you so. At most one saved story travels with any single request.
CueFlect scans text you approve for things that look like names, organisations, email addresses, phone numbers and URLs, and warns you about them — it does not remove them. What you approve is sent as written, so the warning is there to inform the decision rather than to clean up after it.
Other participants' information
When a user starts a session, CueFlect captures the audio their PC is playing, which can include another participant's voice. It transcribes questions and suggests answers on the user's screen. Screenshots the user takes can also contain another person's image, name or information.
Full Local processing stays on that PC. Mixed mode sends bounded text to the gateway, including questions and recent exchanges, but keeps audio and screenshot pixels local. Cloud mode can send screenshot pixels as well as text. Both current editions transcribe speech on the PC by default; selecting the gateway transcription route sends detected speech segments instead. Approved profile text can accompany a request. The mode comparison above shows the boundaries.
CueFlect holds captured audio in memory for transcription and does not write it to a file or offer an audio-retention setting. Sessions explicitly started by the user are saved locally, encrypted for their Windows account, until they delete them. Separately enabled transcript, prompt/answer or screenshot-text diagnostics are stored locally unencrypted; they are off by default and expire after the user's selected retention period of 1–30 days. CueFlect cannot access those local files.
Cloud requests pass through CueFlect's gateway to OpenAI, our inference provider, without account identity added to the inference payload. Speech or free text can still identify someone. The provider may retain content for abuse monitoring or prompt caching under its applicable endpoint and model terms; disabling response storage does not guarantee zero retention. Processing can involve the United States. Contact us for the current retention profile and transfer safeguards. Neither CueFlect nor its inference provider uses this API content to train models under the applicable provider terms. CueFlect does not create voiceprints or identify speakers by voice.
Users must have lawful authority to capture and submit others' information, and must provide any legally required notice and obtain required permissions. This page does not itself notify everyone on a call or replace their consent. CueFlect remains responsible for obligations arising from its own processing. Controller and processor roles depend on the actual activity and applicable law; a person's description of their use as personal or professional does not settle those roles.
For a privacy request, including an objection, write to support@cueflect.com; no CueFlect account is required. Applicable rights can include access, correction, deletion, restriction and objection, and portability where its conditions apply. You can also complain to the relevant supervisory authority. We will assess the request and any provider-held information rather than assume there is nothing to address. For a transcript on another person's PC, that user controls its deletion; we hold no decryption key. Please do not send a full interview transcript just to identify a request.
The company behind CueFlect is planned to be established in Georgia (the country): [TO BE COMPLETED BEFORE LAUNCH: legal name, form, registered address and registration number]. This is a pre-launch candidate disclosure. The activity-specific legal bases, roles, notification arrangements and international-transfer safeguards remain subject to legal review; this notice does not assert that every use is lawful or that notification obligations have been discharged.
Deleting your account
Live-store data is deleted immediately. A small, disclosed set of records is retained, and each one expires or persists according to its own rule in the published deletion and retention matrix — free-allowance eligibility hashes and financial records are the main examples, and each states why.
Measurement
The desktop app asks once, during setup, whether it may send aggregate usage counts, and there is no default answer — until you answer it sends none, and nothing is stored up to send later.
This site asks the same question, in the banner at the bottom of the page. If you agree, your browser sends CueFlect a small set of aggregate events: an event name and a few fixed categories, recorded to the hour it arrived. There is no cookie, no identifier, and no way to join an event to you or to your account — your answer is kept in your browser's own storage, which is never sent with a request. Until you answer, nothing is sent and nothing is held back to send later.
No third-party analytics provider receives CueFlect data. The events go to CueFlect's own service and nowhere else; no analytics vendor's script runs on this site.