Skip to main content
Verbiform
Explore Templates Document styles Pricing Open the app

Privacy Policy

Last updated November 21, 2026

Verbiform turns spoken details into structured records. This policy describes what the product actually does with your data — not what would be convenient to claim. Where information leaves your device, we say so plainly and name who receives it.

The short version

  • Records you create are saved on your device first, and work with no internet connection.
  • Two separate switches control what leaves your device: online processing (transcription and AI) and sync (backup and multi-device copies). Online processing stays off until you turn it on. Sync starts when you sign in, so your records are backed up to your Verbiform account; the Online features switch in Settings turns it off. If you never sign in, your records never leave your device — the one exception is a smartwatch you have paired yourself; see recording from a smartwatch.
  • When sync is on, your records — including audio recordings and photos — are stored on Verbiform’s server so they reach your other devices. So are recordings you saved without choosing a form yet, so you can play them and file them on another of your devices. See recordings you have not filed yet.
  • Choosing “This device only” keeps transcription and AI on your device. It does not turn off sync: while you are signed in, your records are still backed up to your Verbiform account unless Online features is off, and anonymous crash reports are still sent. The one difference is recordings you have not filed yet: a recording made while “This device only” is chosen stays on the device that made it.
  • AI features (auto-fill, summaries, action items, ask) send the text of your notes to an AI provider. In the web app you can instead run AI entirely on your device; the phone app has no on-device AI model.
  • Verbiform transcribes in one of two modes, and you choose which: on this device (the Whisper model runs in your browser or in the phone app; audio goes nowhere) or online, where the audio of a recording is sent to Deepgram, a speech-processing provider, for faster and more accurate results. Online processing is off until you turn it on and accept what it means; every plan includes a monthly allowance of online minutes, and when it runs out the on-device model takes over automatically. For transcription only the audio is sent, not your fields, photos or account details. In the phone app, when online processing is on, the read-aloud voice also sends the text of each spoken reply to Deepgram to turn it into speech. We instruct Deepgram not to use any of this to train models.
  • Speaker labelling (“who said what”) can run entirely on your device, free on every plan. The first run downloads a small model, and after that it works offline with no third party involved. With online processing, labelling rides the same Deepgram call as transcription and counts against the same minute allowance. A Team owner can switch online processing off for their whole workspace, and on-device labelling keeps working when they do.
  • We do not sell your data, and we do not use your records to train AI models.

Who owns your content

Your content is yours. GLE Ventures LLC does not claim ownership of the text, data, voice, photos, or completed documents you create with Verbiform — storing or syncing them so they reach your other devices does not change who they belong to. We use them only to provide the features you turn on, as this policy describes; we do not sell them, and we do not use your records to train AI models. You can export or delete your content at any time.

What stays on your device

Verbiform stores your records in your browser’s local database (IndexedDB) and your preferences in local storage. This includes your notes, form values, contacts, container records, imported PDF templates, and any audio or photos you capture.

Speech-to-text can run entirely on your device. The Whisper model is downloaded to your browser and transcribes locally, so dictation in device mode sends your voice nowhere. It is the mode used whenever you are signed out, offline, have chosen “This device only”, or have used up your plan’s online minutes. In the browser, the same applies to the read-aloud voice and to the optional on-device AI models.

In the phone app, your records are kept in the app’s own local database on the phone, and the Whisper model runs inside the app, so transcription in device mode happens on the phone too. The phone app has no on-device AI model, and its read-aloud voice is the phone’s own voice unless online processing is on.

If you use Verbiform without an account, your records, voice and photos stay on your device — they are not transmitted to us or anyone else. Two narrow exceptions: the first time you use voice or AI features, your device downloads the speech and AI models from a content-delivery network (which sees your IP address and which model you fetched, never your data); and the app may send anonymous, stripped-down crash reports — an error type and code path only, never your content — so we can fix bugs.

Each time it opens, the phone app also asks Expo, our app-update service, whether an update is available. That check shares your IP address and a random install ID, never your content.

What happens when you turn on sync

Sync turns on when you sign in. While it is on, the following is transmitted to and stored on Verbiform’s server:

  • Your records and their field values, including transcripts and notes
  • Audio recordings and photos attached to those records
  • Recordings you have not filed yet: the audio, its transcript and when it was made, kept until you turn it into a record or delete it (see recordings you have not filed yet)
  • Contacts and container records (customers, contacts) you create
  • Workspaces, their membership, and who created or edited each record
  • Your custom form templates — the field definitions, and the raw file of any fillable PDF you imported so it can be filled on your other devices
  • Your branding (logo, colors, business details), form folders and their assignments, “My forms” selections, and your PDF-forms sections
  • Your email address, and your name if your sign-in provider supplies it

Sync is optional and reversible. Turning off Online features (Settings → Devices, sync & backup) stops new syncs; choosing “This device only” does not. You can sign out at any time, and Sign out & remove my data additionally removes the local copies from that device.

Audio recordings and photos are stored encrypted at rest on our server; see security.

Recordings you have not filed yet

When you save a recording without choosing a form, for example with Save for later, it waits on the device under “Needs a form”. While you are signed in, Verbiform also keeps a copy on your Verbiform account, so you can play it and turn it into a record on another of your signed-in devices. A lost phone or a cleared browser then does not lose it.

  • What is kept: the audio, the transcript, when it was made and the time zone, which device made it (for example “Chrome on Windows”), and the workspace it was made in.
  • Only you can see it. It is not shared with a team. Owners and admins of a shared workspace never see the recordings you have not filed.
  • What stays on your device instead:
    • a recording made while “This device only” is chosen, at the start or at the end of the recording;
    • a recording in a workspace that has a medical or healthcare form, or any record that stays on the device (see health data);
    • a recording whose workspace or form your device could not check at the time;
    • a recording with photos, and one larger than 25 MB.
    You can choose Back up to your Verbiform account for a recording that was kept on your device for the first or third reason once your device can check it again, and only while “This device only” is off. It is never offered for a recording in a workspace with a health form.
  • When it is taken back off your account: if a medical or healthcare form later appears in that workspace on any of your devices, or you switch the device that made it to “This device only”, the copy on your account is deleted. The recording stays on the device that made it.
  • When it is deleted: the copy on your account is deleted when you turn the recording into a record (the record then syncs like any other), when you delete the recording on any of your devices (it is then deleted from all of them), or when it is taken back off as described above. A deleted recording that was on your account is hidden and kept on a device for 7 days before it is removed there, in case it was deleted by mistake on another device.
  • What we keep after that: a small marker with the recording’s id and how it ended (filed, deleted or taken back off), for up to 400 days, so a device that was offline for a long time cannot bring the recording back. The marker holds no audio, no transcript and no device name.
  • Storage: the audio counts toward the same account storage allowance as the audio and photos on your records, and like them it is stored encrypted at rest on our server.
  • Processing: keeping a recording on your account never sends it to a transcription or AI provider. When you turn it into a record, transcription and AI run on the device you are using, under its usual settings; a recording made while “This device only” was chosen is transcribed on that device.

Recording from a smartwatch

A smartwatch cannot transcribe. If you pair one, its recordings travel to your phone or computer through Verbiform’s server, which holds the audio only long enough for one of your own signed-in devices to collect it. This is the one path where audio reaches our server without you turning on sync, so it is worth being exact about:

  • The audio is uploaded to our server as soon as the watch has network, whether or not record sync is switched on. Pairing a watch is what enables this; unpairing it stops it.
  • It is encrypted in transit (HTTPS) and stored encrypted at rest on our server. It is not end-to-end encrypted — we hold it in a form we could technically read, and we are telling you that rather than implying otherwise.
  • We never transcribe it and never send it to an AI provider. The server stores bytes and status; the transcription happens on your own device.
  • The server copy is deleted the moment one of your devices turns the recording into a record — usually seconds after you open the app.
  • Audio nobody collects is deleted after 14 days, whether or not it was ever processed. The recording on your watch is not affected — it stays there until a record is made from it.
  • Removing a watch deletes any of its audio still waiting on our server, immediately. So does deleting the recording from the app, and so does deleting your account.
  • The copy on the watch is deleted once the watch is told a record exists. Until then the watch keeps it, so a failed transfer never loses your recording.

If you never pair a watch, none of this applies to you and no audio is relayed.

Who else processes your data

We use a small number of third-party services. This list is kept current; if it changes, this page changes with it.

ProviderWhat it receivesWhen
DigitalOcean Hosting for everything above — the server and database live here (United States) Whenever sync is on
Model delivery (Hugging Face / jsDelivr) Your IP address and which speech / AI model your device is downloading — never your records, voice or notes The first time you use voice or on-device AI, then cached
Stripe Your name, email and payment details — entered directly with Stripe; card numbers never touch our server Only when you subscribe or manage billing
Deepgram The audio of a recording being transcribed or speaker-labelled online and, in the phone app, the text of a reply read aloud by the online voice while online processing is on; never your fields, photos or account details. We instruct Deepgram not to use it to train models Only when online transcription is on, within your plan’s monthly allowance
Google (Gemini) The text of the note or transcript being processed — not audio When you use an AI feature with online AI selected
Google / Microsoft / Apple Your email address and name, to confirm who you are Only if you sign in with that provider
Expo (app updates) Your IP address, the phone’s platform, the app version and a random install ID, to check for an app update — never your records Phone app only, each time it opens
cdnjs (code library delivery) Your IP address, while the phone app downloads the PDF reader library (pdf.js) — never your records or files Phone app only, when it shows or reads a PDF (a document preview, mapping a flat PDF, importing a PDF syllabus); usually downloaded once, then cached
Google (ML Kit text recognition) Device and app information and usage measurements, such as the phone model, the app version, how long recognition took and any error codes — never the photo or the text in it, which are read on the phone Phone app only, when you read text from a photo
Google (STUN) The IP addresses of the two devices, so they can find each other — the records travel directly between the devices, end-to-end encrypted Only during a device-to-device transfer
Email delivery Recipient address and message content for workspace invitations and any record you choose to send Only when you send something

You can avoid the AI provider entirely. Verbiform can run its AI features on a model downloaded to your device, in which case no note text is transmitted for AI processing. To choose this, pick “This device only” under Settings → Privacy. The on-device model is managed under Settings → Advanced & integrations.

The phone app has no on-device AI model. There, “This device only” fills fields without AI, and features that need the AI provider are not available.

How AI features use your content

AI is used to draft field values from what you said, summarise a note, extract action items, and answer questions about your own records. Only the content needed for that request is sent.

Your content is not used to train AI models — neither ours nor our providers’. We do not sell your data or share it with advertisers.

AI output can be wrong. Drafted field values are marked with a confidence level and the exact words they came from, and are meant to be reviewed before you rely on them.

Sharing and workspaces

If you share a workspace, the people you invite can see the records in it, and — if you give them the editor role — change them. Edits are attributed to the account that made them. Removing someone’s access stops future syncing, but does not retrieve copies already downloaded to their device.

Recordings someone has not filed yet are private to them, even in a shared workspace: no one else in the workspace, including its owner, can see them. A person removed from a workspace keeps their own unfiled recordings, and can file them into another workspace.

Retention and deletion

Deleted records go to Trash and are recoverable for 30 days, after which they are removed. Deleting a record locally also removes it from the server on your next sync. A recording you have not filed yet is deleted from your account when you file it, delete it, or it is taken back off; see recordings you have not filed yet for the small marker we keep for up to 400 days.

You can export a complete copy of your data from Settings at any time. To delete your account and everything stored for it, use Delete account on the Account page (Settings → Devices, sync & backup → Manage account & sign-in): it erases all server-side data immediately, cancels any active subscription, and wipes records from that device. This cannot be undone, so export a backup first if you want to keep a copy.

Security

All traffic between the app and our server uses HTTPS. Passwords are stored only as salted hashes, and session tokens are stored hashed, so a copy of our database does not reveal either. Audio recordings and photos stored on our server, including recordings you have not filed yet and smartwatch recordings on their way to your phone, are encrypted at rest (AES-256), with the key kept apart from the database, so a copy of the database or of its nightly backup does not reveal them either. This is not end-to-end encryption: our server decrypts them to send them to your own signed-in devices. The database is backed up nightly.

For each device signed in to your account, we keep the device type and an approximate location (city and region) worked out from its internet address, so you can recognise your devices and spot sign-ins that are not you; we do not keep the internet address itself. The location is looked up on our own server, so your internet address is not sent to anyone else for this. IP geolocation by DB-IP.

You can add a PIN or biometric lock on a device. No system is perfectly secure, and we will tell you promptly if a breach affects your data.

Health data

Verbiform is a general business tool and is not currently designed for protected health information (PHI) or other regulated health records. A dedicated, HIPAA-ready healthcare edition is on our roadmap — until it is available, please keep clinical and regulated health data out of Verbiform.

Medical and healthcare records, such as patient and clinical forms, always stay on your device: they are never synced to your account or sent for online processing. Recordings you have not filed yet follow the same rule by workspace: in a workspace that has a medical or healthcare form, or a form with a field marked as private health information, they stay on the device that made them, and if such a form appears later, any copy already on your account is deleted. Other forms that ask about health, such as massage intake, tattoo consent, fitness assessments and wellness session notes, are handled like any other record: health details typed into them sync to your account and can be processed online when online processing is on. Please do not back up a recording that contains patient health information.

Children

Verbiform is not intended for children under 13. We do not knowingly collect personal information from children under 13. If you are a parent or guardian and believe a child under 13 has provided personal information, contact support@verbiform.com and we will delete it.

Changes to this policy

If we change how data is handled in a way that matters (for example, moving speech-to-text off your device to an online service), we will update this page and tell you in the app before it takes effect. Changes that expand what leaves your device will be something you opt into, not something that happens automatically.

Contact

Questions, data requests, or account deletion: support@verbiform.com

On this page

The short version Who owns your content What stays on your device What happens when you turn on sync Recordings you have not filed yet Recording from a smartwatch Who else processes your data How AI features use your content Sharing and workspaces Retention and deletion Security Health data Children Changes to this policy Contact
Verbiform

Talk it through; Verbiform turns it into a document you can check, sign, and send. Made for work, study, and everyday life. Developed & Designed in Texas, USA.

Product

How it works Templates Document styles Pricing

Who it’s for

Office & meetings Field service & trades Inspections & reports Students Teaching Personal

Company

Support Contact sales Your data Trust & accuracy

Legal

Terms of Service Privacy Policy Refund Policy
© 2026 GLE Ventures LLC. All rights reserved. Verbiform™ is a wholly owned brand of GLE Ventures LLC. Voice in. Document out.