Privacy Policy for Local Flow

Last updated / Effective date: 26 June 2026

App: Local Flow (Android application id: dev.dictation.app)

Developer / Data controller: Pasquale D'Alterio

Contact: hello@modelscout.co

The short version

Local Flow is an on-device dictation app. Your voice and the text you dictate are processed entirely on your phone and never leave your device. There is no cloud transcription, no server that receives your audio or your dictated text, and no account is required to use the core dictation feature.

The App does use the internet for one thing only: a one-time download of the language models it needs to run (see Section 6). That download transfers model files to your device; it never sends your voice or your dictated text anywhere. After the models are downloaded, dictation works on-device.

We designed Local Flow this way on purpose: privacy is the whole point of the product. The rest of this policy explains exactly what that means, what limited data is involved, and what your choices are.

1. Who we are

Local Flow is published by Pasquale D'Alterio, an individual developer ("we", "us", "the developer"). For the purposes of data-protection law, the developer is the data controller for the limited data described in this policy. You can contact us at hello@modelscout.co.

2. How Local Flow works (and why this matters for your privacy)

When you dictate with Local Flow:

1. The app captures audio from your microphone only while you are actively dictating.
2. That audio is converted into a raw transcript by a speech-recognition (ASR) model that runs locally on your device.
3. The raw transcript is cleaned up (punctuation, casing, fixing recognition errors and disfluencies) by a second model that also runs locally on your device.
4. The cleaned text is inserted into the text field you are typing in.

All of these steps happen on your phone. The audio and the text are processed in your device's memory and are not sent to us, not sent to any server, and not sent to any third party.

3. Data we do NOT collect

To be explicit, Local Flow does not:

• Collect, transmit, store, or sell your voice audio. Audio is processed in memory on your device for the purpose of recognition and is not retained as a file or uploaded anywhere.
• Collect, transmit, store, or sell the text you dictate (your transcripts). Dictated text is produced on your device and inserted into the app you are using; we never receive it.
• Require you to create an account or provide your name, email, or phone number to use core dictation.
• Use your microphone for any purpose other than dictation, or while you are not actively dictating.
• Use the accessibility service to read, harvest, log, or transmit the contents of your screen or the apps you use (see Section 5).
• Track your location.
• Build advertising profiles about you or share data with advertising networks.

Note: the App does request internet access (it needs it to download the language models once — see Section 6). Internet access is not used to send your voice or dictated text anywhere; it is used only to fetch the model files and for the App's normal functioning on Google Play. The statements above describe the shipped build; they are kept consistent with the Google Play Data Safety form for this App.

4. The microphone permission

What it is: Permission to access your device's microphone.

Why we need it: To hear you when you dictate. This is essential to the app's function.

How we use it: The microphone is activated only when you start a dictation and is released when you stop. Audio is converted to text on your device and is not recorded to a file, saved, or transmitted off the device by Local Flow.

You can revoke the microphone permission at any time in Android Settings, but dictation will not work without it.

5. The accessibility service permission (please read — we know this one matters)

Local Flow uses Android's Accessibility Service. We understand that accessibility permissions are powerful and that users (and Google) rightly scrutinize apps that request them. Here is exactly what we do and do not do with it.

Why Local Flow needs it: The accessibility service is the mechanism Android provides that lets Local Flow insert your dictated text into the text field of whatever app you are using, at your cursor position. Without it, the app could not place cleaned-up text into other apps' text boxes — which is the core purpose of a dictation tool.

What we use it for — and ONLY this:

• Detecting that there is an editable text field in focus, so we know where to place your dictated text.
• Inserting (and, where you ask, editing around the cursor of) the text you dictated into that field.

What we explicitly do NOT do with the accessibility service:

• We do not read, collect, log, store, or transmit the existing contents of your screen, your messages, your passwords, your browsing, or the apps you use.
• We do not monitor your activity in other apps for analytics, advertising, or any purpose unrelated to inserting your dictated text.
• We do not send any accessibility-derived information off your device.

In short: the accessibility service is used as a "typing" mechanism to deliver your own dictated words into the field you chose, and for nothing else. All of this happens on your device.

You can turn the accessibility service off at any time in Android Settings > Accessibility, but text insertion into other apps will not work without it.

This description matches what the App actually does: the accessibility service reads the currently focused text field (so dictated text lands at the correct cursor position) and the package name of the app in the foreground (so you can disable the bubble per app). It does not read other on-screen content, and nothing derived from the accessibility service leaves your device. The same description is provided as a prominent in-app disclosure shown before you enable the service, and in the accessibility-use declaration for this App in the Google Play Console.

6. One-time model download

Local Flow needs two machine-learning models (a speech-recognition model and a text-cleanup model) to work. These models are downloaded once to your device (approximately 1 GB total) and then run locally. They are not re-downloaded each time you dictate.

• What is downloaded: the model files needed for on-device speech recognition and text cleanup.
• From where: the models are downloaded from the model-hosting location operated for Local Flow. The current host is identified on the Google Play listing and may be a content-delivery network or Google Play's own asset-delivery mechanism; in either case it serves only the model files.
• What is sent during download: downloading a file necessarily reveals standard network information to the host serving it (such as your IP address and a request for the file), as with any download. This is a normal technical consequence of fetching a file. The download does not include any of your voice or dictated text — it is a one-way transfer of model files to your device.

After the one-time download, dictation works on-device and does not require the model host.

7. Payments and the Pro subscription

Local Flow offers a free tier and a paid "Pro" tier (monthly, yearly, or lifetime). All purchases are processed by Google Play Billing. We do not receive, handle, or store your payment card number or other payment details.

When you buy Pro, Google processes your payment under Google's own terms and privacy policy. We receive only the limited purchase/entitlement information that Google Play provides to developers (for example, confirmation that a valid purchase or subscription exists) so that the app can unlock Pro features. We do not use this to identify you beyond managing your entitlement.

• Google Play Terms of Service: https://play.google.com/intl/en/about/play-terms/
• Google Privacy Policy: https://policies.google.com/privacy

For questions about charges, refunds, or cancellations, see Section 9 and your Google Play account.

8. Analytics and crash reporting

We aim to collect as little as possible. Local Flow does not collect your voice or dictated text for analytics, crash reporting, or any other purpose.

The app is distributed through Google Play and may rely on standard Google Play services. As a result, Google may collect certain aggregate, non-content technical and diagnostic information (for example, install/uninstall statistics, crash and performance data, and general device information) under Google's own policies. Any such information does not include the content of your dictation.

The shipped build includes no third-party analytics or crash-reporting SDK — for example, there is no Firebase, no Crashlytics, no Sentry, and no advertising SDK. The only diagnostic information that may be collected is the standard, aggregate crash and performance data that Google Play itself gathers for any app (Android vitals), under Google's own policies. None of that information includes the content of your dictation.

9. Data retention and your control

Because Local Flow processes your voice and text on your device only:

• Your voice audio is not retained after a dictation is processed.
• Your dictated text is delivered into the app you were using; from that point it lives in that app, under that app's control, not ours.
• Any settings, preferences, and your personal dictionary (the custom terms you add) are stored locally on your device. They are not uploaded to us.
• Dictation history: Local Flow keeps a history of your past dictations on your device so you can review them. This history is stored only on your device and is encrypted at rest (using a key held in the Android Keystore). It is never transmitted to us or any third party, and remains under your control — you can clear it in the app, and uninstalling the app deletes it.
• Off-device backup is disabled. The app sets android:allowBackup="false" and android:fullBackupContent="false", so your app data (personal dictionary, settings, and dictation history) is not included in Android's system/cloud Auto Backup and is not extractable by adb backup. This closes the off-device copy path for that data.
• Uninstalling the app removes all of the app's local data, including the personal dictionary, dictation history, and the downloaded models.

You remain in control: you can clear app data, revoke permissions, or uninstall at any time through Android Settings.

10. Your rights

Depending on where you live (for example, under the UK GDPR / EU GDPR or similar laws), you may have rights to access, correct, delete, or restrict the processing of personal data we hold about you, and to object to certain processing.

Because we do not collect your voice or dictated text, and core dictation requires no account, in practice we hold very little (if any) personal data about you. If you believe we hold personal data about you and wish to exercise any of these rights, contact us at hello@modelscout.co and we will respond as required by applicable law. For payment-related data, contact Google (see Section 7). You also have the right to lodge a complaint with your local data-protection authority. The developer is based in the United Kingdom; the relevant supervisory authority is the UK Information Commissioner's Office (ICO, https://ico.org.uk), consistent with the EULA's choice of the laws of England and Wales. If you are in the EU/EEA, you may instead contact your local data-protection authority.

11. Children's privacy

Local Flow is not directed at children and is not intended for use by children under the age of 13 (or the higher age of digital consent that applies in your country). We do not knowingly collect personal data from children. If you believe a child has provided us with personal data, contact us at hello@modelscout.co and we will take appropriate steps. The App's content rating is set in the Google Play Console for its target markets.

12. Changes to this policy

We may update this Privacy Policy from time to time. When we do, we will revise the "Last updated" date at the top and, where appropriate, provide notice in the app or on the listing. Your continued use of Local Flow after an update means you accept the revised policy.

13. Contact us

If you have any questions about this Privacy Policy or about how Local Flow handles data, contact:

Pasquale D'Alterio
hello@modelscout.co