Privacy Policy

Last updated 13 September 2026

ScribeLLM is operated by Garg Capital Group LLC. This policy explains what the app collects, who it is shared with, and how to get rid of it. It describes what the system actually does.

What we collect

Category	What it is	Why
Account	Name, email address, password (stored only as a hash).	To identify you and send your notes.
Audio recordings	The conversation audio you record.	To produce a transcript and summary.
Transcripts and notes	Text derived from your recordings.	To show you the result and email it.
Participant names	Names you optionally type in.	To label who said what.
Diagnostics	Crash reports, error messages and timing information, plus the device model, operating system version, app version and a random installation identifier.	To find and fix failures, and to know which app versions are affected. Never includes transcript text, recordings, patient details, names or email addresses.
Usage	Which screens you open, which actions you take, and how long each request to our server takes.	To see which parts of the app are slow or failing. Never includes what you recorded, typed or were shown.
Network address	The IP address your device connects from, and the app version and platform it sends.	To operate and secure the service: to see where traffic comes from, to investigate failures, and to detect abuse. Kept with the diagnostic records and deleted on the same schedule.

We do not collect contacts, precise location, advertising identifiers, or browsing activity, and we show no advertising. We do not sell personal information, and we do not share it for cross-context behavioural advertising.

Diagnostics, usage and your IP address

ScribeLLM records operational information about the app and the service so that failures can be found and fixed. This is separate from your recordings and notes, and it is held separately.

Crash reports. When the app fails, it stores the error and the technical trace that produced it, along with the screen you were on and the last few actions you took — recorded as short labels such as “opened settings”, never as the content of what you typed or were shown. The report is uploaded the next time the app starts.
Usage. Which screens are opened and which actions are used, so we can see what is slow or failing. Not the content of any screen.
Performance. How long each request to our server takes, which endpoint it was, and whether it succeeded.
Your IP address. Every request to our server is recorded with the network address it came from, together with your app version and platform. An IP address is personal information in some jurisdictions, so we say so plainly: we keep it to operate and secure the service — to investigate failures and detect abuse — and for no other purpose. It is never used to build a profile of you, never combined with your recordings, and never shared for advertising.
A device identifier. A random identifier generated by the app on your device, so that crashes from the same device can be grouped. It is not your advertising identifier, and it is regenerated if you reinstall the app.

This information is processed on our own servers. It is not sent to any analytics or crash-reporting company, and no third-party analytics SDK is present in the app.

Diagnostic, usage and network records are deleted automatically after 90 days. Crash reports are kept for 180 days, because a fault that appears rarely needs longer than one window to be recognised. Aggregated counts with no identifiers in them may be kept longer.

Who else processes it

Producing notes from audio requires specialist services. Each receives only what it needs:

Provider	Location	What it receives
Amazon Web Services	United States	Stores audio recordings, transcripts and account data. Encrypted at rest and in transit.
Deepgram	United States	Converts recorded audio to text and separates speakers. Receives the audio.
OpenAI (ChatGPT)	United States	Generates the summary sections you select. Receives the transcript text, not the audio.
Google (Gemini)	United States	Generates the summary sections you select. Receives the transcript text, not the audio.
Anthropic (Claude)	United States	Generates the summary sections you select. Receives the transcript text, not the audio.
Resend	United States	Delivers summary emails you choose to send. Receives the notes in the email body.

Cross-border transfer. These providers process data in the United States. If you are in the EEA, UK or India, using ScribeLLM means your recordings are transferred there for processing.

Automated processing and accuracy

Which AI services process your conversations

ScribeLLM cannot work without sending your data to third-party AI providers. Naming them is the point of this section — “we use AI” tells you nothing about where your conversation actually goes.

Deepgram (United States) receives the audio of every recording. It produces the transcript and works out who spoke when. This is the only provider that receives audio.
OpenAI (ChatGPT), Google (Gemini) and Anthropic (Claude) (all United States) receive the transcript text — never the audio — and write the summary sections you selected. You choose which of these models writes your notes, in the app’s settings or on the screen shown after a recording. Choosing more than one sends the same transcript to each, so their answers can be compared.

These providers process your data only to return a result to us. We do not permit them to use your recordings, transcripts or notes to train their models, and we have no arrangement under which they may do so. We do not sell your data or use it for advertising.

The full list of every processor, including storage and email, is in the table below.

Accuracy

Transcription and summarisation are automated. They make mistakes: speakers can be misattributed, words misheard, and a summary can omit or distort something that was said. Notes are a draft to review, not a record of what happened. Do not rely on them for anything consequential without checking the audio, which the app keeps for exactly that reason.

How long we keep it

Recordings, transcripts and notes are kept until you delete them. Deleting a conversation removes its audio, transcript and notes. Deleting your account removes everything, including the audio files themselves. Neither is recoverable.

Sent emails are held briefly by our email provider for delivery, per its own retention.

Security

Data is encrypted in transit (TLS) and at rest (AES-256). Recordings are stored in a private bucket with no public access. Passwords are stored only as bcrypt hashes and are never recoverable, by us or anyone else.

No system is perfectly secure. We do not claim otherwise.

Your rights

You can access, export, correct or delete your data at any time from within the app. Deletion is immediate and permanent. If you are covered by the GDPR, the UK GDPR, India's DPDP Act or the CCPA, you also have rights of access, rectification, erasure, portability and objection — write to support@chaalbaazi.com and we will respond within 30 days.

Children

ScribeLLM is not directed at children under 13 and we do not knowingly collect their data.

Other people in your recordings

Your recordings contain other people's voices and words. You are the controller of that content; we process it on your instruction. It is your responsibility to have their permission before recording. If someone asks to have a recording of them deleted, delete that conversation or contact us.

Changes

If this policy changes materially we will say so in the app before the change applies.

Contact

Garg Capital Group LLC — support@chaalbaazi.com