Signalment

Privacy Policy

What we hold,
and what we do with it.

This policy covers signalment.ai, the Signalment web application, and the Signalment recorder app for phones. Every company that touches your data is named below, along with how long each kind of data is kept.

Effective 7 September 2026 · Last updated 7 September 2026

The short version

What we hold
Your email and name from Google, the exam notes you paste or dictate, the records we draft from them, the client and patient names you file them under, audio from recorded visits, and your subscription details.
What we never do
We do not sell your personal information. We do not share it for cross-context behavioural advertising. We do not advertise on your data, and we never show one clinic another clinic's cases.
What we do do
We use your content to run the product and, as section 5 explains in full, to improve it — including training and evaluating AI models. If that is not acceptable to you, do not use Signalment.
Audio
Kept only while we need it. Today that means it goes once the transcript exists — but we are not promising a fixed lifetime, and §8 says what we currently do. Transcripts and their records expire after 30 days whether you draft from them or not.
Who else sees it
Amazon, Anthropic, Deepgram, Google, Stripe and the National Library of Medicine. All process data in the United States.
Getting it back or out
Delete any record from the app at any time. Delete your whole account from the account page and the notes go with it. Or email us and we will do it.

Where this summary and the full text differ, the full text governs.

01 Who we are and what this covers

Signalment is operated by Signalment, LLC, a United States limited liability company. In this policy, "Signalment", "we" and "us" mean that company; "you" means the veterinary professional or practice using the service.

This policy applies to the marketing site at signalment.ai, the web application at signalment.ai/app, the Signalment recorder app for Android and iOS, and the emails we send you. It does not apply to anything a third party does on its own site, including your practice-management software, Google, or Stripe.

For the information you and your practice give us about yourselves, we are the controller: we decide what happens to it. For the clinical content you put into Signalment, we act on your instructions; section 3 explains what that means for the people and animals in your records.

02 What we collect

Your account

Signalment signs you in with Google and nothing else. From that sign-in we receive and store your Google account identifier, your email address, whether Google has verified it, and, when Google supplies them, your name and profile picture. We never receive or store your Google password, and we do not read your Gmail, Drive, Calendar or contacts.

What you put into the product

  • Exam notes. Whatever you type, paste or dictate into the drafting box.
  • Drafted records. The SOAP note, differentials, plan, gap check and client letter the model produces, along with which model produced them.
  • Filing names. The client's name and the patient's name you file a record under, as you type them.
  • Audio and transcripts. If you use the recorder app or the in-app recorder: the audio of the visit, its length and start time, the transcript, and your confirmation that the client agreed to be recorded.
  • Literature lookups. The search PubMed was asked and the abstracts it returned.

Billing

Subscriptions run on Stripe. Stripe collects and holds your card details; we never see or store a full card number. We store your Stripe customer identifier, which plan you are on, its status and renewal date, whether a trial has been used, the billing name and email you give us, and the invoice history Stripe reports back. When Stripe Tax is enabled, Stripe also collects the billing address needed to calculate tax.

Usage

We record how you move through the product: sign-ups, notes drafted, whether a draft succeeded or failed, plan pages viewed, checkouts begun, trials started, subscriptions changed, and how many notes and recordings you have used against your allowance. We also log ordinary technical data — IP address, browser and device type, timestamps, error traces — which our infrastructure produces as a matter of course.

Product analytics events carry no clinical content of any kind. There is no path in our code by which a note's text, its title, a client's name or a patient's name can reach an analytics system, and your account is identified in analytics only by an irreversible salted hash. The restriction is enforced in the code that sends the events.

What you send us

Emails you send us, and anything you choose to put in them. If you report a problem with a draft and paste the draft in, we will have that draft.

What we do not collect

We do not ask for or want your DEA number, your licence number, your practice's financial records, or your clients' payment information. We do not track your location. We have no advertising trackers, no social media pixels, and no session-replay tooling on the site or in the apps.

03 Your clients and their animals

A veterinary record carries information about the owner as well as the animal: the name it is filed under, and, on a recording, their voice. Signalment holds that information because you put it there.

You decide what goes in. We process client information on your instructions and for the purposes you use the product for. You determine what to record, what to type, what to file a record under, and how long to keep it. Your obligations to your clients under your state's law, your practice act and your own privacy notice are yours; we do not take them on, and using Signalment does not discharge them.

Consent to record is your responsibility. Twelve US states require every party to a conversation to agree before it may be recorded, and California is one of them. The recorder app will not start until you confirm the client agreed, and our servers reject a recording submitted without that confirmation — but confirming it in the app is not the same as obtaining it in the room. Obtaining it, and being able to show you obtained it, is on you. The Terms of Service set this out in full.

Keep names out of the notes. The product asks for the owner's name in a separate field precisely so that it does not have to sit inside the clinical text. Put it there. Beyond names and what you dictate, we do not want and do not ask for your clients' addresses, phone numbers, payment details or government identifiers.

Isolation. Records are stored under your account and are readable only by you. No clinic is ever shown another clinic's cases.

Requests from your clients. If one of your clients contacts us directly about information in your records, we will not act on it ourselves — we will tell them to contact you and, where we can identify the account, let you know.

HIPAA does not apply. Animal health records are not protected health information, so HIPAA imposes no obligations here and we make no HIPAA claim. If you also handle human health information, Signalment is not the place for it.

04 How we use it

  • To run the service: signing you in, drafting records, transcribing recordings, looking up literature, storing and searching your notes, and counting them against your allowance.
  • To bill you: starting trials, taking subscription payments through Stripe, handling failed payments, pauses, plan changes and cancellations, and issuing invoices.
  • To write to you: a welcome message, a warning three days before a trial ends, a note when you reach your allowance, a short series of onboarding emails in your first ten days, and service notices about outages, security and changes to these documents. You can stop the marketing and onboarding ones; you cannot stop the ones about your own account and its security.
  • To understand and improve the product: measuring which steps people complete and where they stop, and the uses set out in section 5.
  • To keep it working and safe: debugging, monitoring, capacity planning, and detecting abuse, fraud and attempts to break in.
  • To meet legal obligations: tax and accounting records, responding to lawful requests, and establishing or defending legal claims.

We do not make decisions about you by automated means that produce legal or similarly significant effects.

05 Improving Signalment, and training AI models

Read this section before you put a real case into Signalment. Under our Terms of Service you grant us a broad licence to use the content you submit, and that licence includes developing and training AI models. If you are not comfortable with that, do not use the product.

We use the content you submit — your exam notes, the records we draft, transcripts, and the corrections and edits you make — to operate, debug, evaluate and improve Signalment and to develop new features and products. That includes training, fine-tuning, evaluating and testing AI models, our own and those we build on.

Two limits apply:

  • Audio recordings are not used for biometric purposes. We do not create voiceprints, we do not use recordings to identify or authenticate any speaker, and we do not use them to infer anything about a speaker's characteristics. This holds however long we keep the audio.
  • Nothing is ever shown to another customer. No output of this work will reproduce your content, your clients or your cases to another clinic.

Separately, the companies whose models we call are bound by their own agreements with us. Anthropic does not train its models on the content we send through its API. Where a provider's default terms would allow training on our traffic, we opt out where that option is offered.

If you would rather your content were not used to improve the product, email us at hello@signalment.ai. We will tell you what we can exclude and what has already been used, and confirm in writing what we have done.

06 Who else touches it

These are all of our subprocessors:

CompanyWhat it doesWhat reaches it
Amazon Web ServicesHosting, storage, databases, email delivery (US East)Everything — Signalment runs on AWS.
AnthropicThe Claude models that draft records and build search queriesYour exam notes and the record being drafted.
DeepgramSpeech-to-text for recorded visitsThe audio of a recorded visit, fetched once from a short-lived link.
GoogleSign-in, and Google Analytics on the site and appYour account identity for sign-in; for analytics, only opaque identifiers and event counts.
StripeSubscriptions, card payments, taxYour billing name, email, address and card details, which Stripe collects directly.
US National Library of MedicinePubMed literature searchesA search query built from the assessment section of a record, plus our contact address. No client or patient name is included.

Each is bound by contract to use what it receives only to provide its service to us. We will keep this table current; a material addition will be announced as described in section 14.

Beyond those, we may disclose information:

  • To professional advisers — lawyers, accountants, auditors and insurers — under duties of confidentiality.
  • When the law requires it, or to establish, exercise or defend legal claims. We will tell you before we hand over your content in response to a legal demand unless we are legally prohibited from doing so, and we will push back on demands that are overbroad or defective.
  • In an emergency, where we believe in good faith that disclosure is necessary to prevent imminent physical harm.
  • In a corporate transaction — a merger, financing, acquisition or sale of assets — in which case the acquirer takes the data subject to this policy, and we will tell you before it becomes subject to a materially different one.

We do not sell personal information, and we do not share it for cross-context behavioural advertising, as those terms are defined under California law and its equivalents. We have not done so in the preceding twelve months.

07 Cookies and analytics

The site and the web app use Google Analytics 4 to measure how the product is found and used. The API also reports a small number of events to Google Analytics directly, so that a subscription started days after a visit is counted against the visit that started it, and keeps a first-party log of the same events in our own database.

What that means in practice:

  • Advertising storage, ad personalisation and ad user data are switched off everywhere, for everyone. We do not advertise on this data and Google is not permitted to use it for advertising signals.
  • In the EEA, the UK and Switzerland, analytics storage is denied by default. Until we ship a consent banner, visitors from those countries are measured without cookies or identifiers.
  • Your account appears in analytics as an irreversible salted hash of your account identifier — never your email, your name, or your Google identifier.
  • Our own first-party event log carries no clinical content and expires after 400 days.

The web and recorder apps also use your browser's or device's local storage to keep you signed in and to hold recordings that have not finished uploading. That is not tracking, and clearing it signs you out.

Most browsers offer a Do Not Track setting; there is no agreed standard for what a site should do with it, and we do not act on it. We do honour Global Privacy Control signals, though since we neither sell nor share personal information there is nothing for such a signal to switch off.

08 How long we keep it

WhatHow long
Audio of a recorded visitKept while it is needed to produce your transcript and to check and improve transcription quality. As things are configured today it is deleted once the transcript exists — or, where your plan has no transcription room, once it does — and storage removes anything still there at 31 days. This is a description of current practice, not a fixed commitment: if we start keeping audio for longer we will say so here, and §14 is how you would hear about it.
Transcript and its recording entry30 days from the recording, whether or not you drafted from it. A record you drafted keeps its own copy of the text.
Drafted records and their exam notesUntil you delete them, or delete your account.
Account detailsUntil you delete the account.
After account deletionYour records are deleted. We keep a minimal marker — the account identifier, the date, and the fact a trial was used — so that a deleted account cannot be recreated to take a second free trial.
Product analytics events400 days, then automatically deleted.
Billing and tax recordsAs long as tax, accounting and audit law requires, generally seven years. Most of this sits with Stripe.
Email delivery logs, server and security logsShort operational periods, typically under 90 days.
De-identified and aggregated dataIndefinitely. It is no longer about an identifiable person.

Deletion from live systems is immediate. Encrypted backups roll off on their own schedule, so a deleted item can persist in a backup for a short period; we do not restore deleted content from backups except to recover from a failure of the whole system.

One caution about your own obligations: Signalment is not your system of record. Your practice's recordkeeping duties run to your practice-management system, and the retention periods above are ours, not a substitute for yours.

09 Security

Everything is encrypted in transit with TLS and encrypted at rest. Audio sits in a private bucket with all public access blocked, reachable only through short-lived signed links. Each request is authenticated against your Google identity, and every query for records is scoped to your account, so one account cannot read another's. Our deployment pipeline holds no long-lived cloud credentials, secrets live in AWS Systems Manager rather than in code, and every change is scanned for leaked secrets before it can ship.

We hold no SOC 2 report and no ISO 27001 certificate.

No system is perfectly secure, and we cannot guarantee that ours will not be breached. If a breach affects your personal information we will tell you and any regulator that must be told, without undue delay. Use a Google account with two-factor authentication, and tell us at once if you think your account has been used by someone else.

10 Your choices

  • Delete a record. Any record, at any time, from the app.
  • Delete your account. From the account page. Your records go with it. This cannot be undone.
  • Get a copy. Every record can be copied out of the app as plain text. For a machine-readable export of everything we hold, email us.
  • Correct something. Filing names and notes are editable in the app; for anything else, email us.
  • Stop the emails. Unsubscribe from any marketing or onboarding email, or email us. Account, billing and security notices continue.
  • Discard a recording. Before it uploads, from the recorder app. After it uploads, dismiss it in the web app or let it expire.
  • Object to your content being used to improve the product. Email us, as described in section 5.

Email hello@signalment.ai for any of these. We will verify that the request comes from the account holder — usually by requiring it to come from the account's email address — and we will respond within 45 days, and normally much sooner. We will not charge you, discriminate against you, or degrade your service for exercising a privacy right.

11 US state privacy rights

Residents of California, Colorado, Connecticut, Delaware, Florida, Indiana, Iowa, Kentucky, Maryland, Minnesota, Montana, Nebraska, New Hampshire, New Jersey, Oregon, Rhode Island, Tennessee, Texas, Utah and Virginia have rights under their state's privacy law: to know what is collected and why, to access a copy, to correct it, to delete it, to obtain it in a portable form, and to opt out of sale, of sharing for targeted advertising, and of certain profiling.

We give every user of Signalment all of those rights, wherever they live, through section 10. There is nothing to opt out of on sale, sharing or targeted advertising, because we do none of them. We do not use or disclose sensitive personal information for any purpose that would require an opt-out, and we do not knowingly process the personal information of anyone under 16.

California. The categories in section 2 correspond to the CCPA categories of identifiers, commercial information, internet activity, audio information, professional information, and inferences; the sources, purposes and recipients are in sections 2, 4 and 6, and retention is in section 8. California residents may designate an authorised agent to act for them; we will ask for proof of the authorisation and to verify the resident's identity. Under California's "Shine the Light" law we disclose no personal information to third parties for their own direct marketing.

If we decline a request we will say why. If you disagree, you may appeal by replying to our decision; we will review it and respond in writing within 45 days. If we get it wrong you can complain to your state attorney general, and Californians may also contact the California Privacy Protection Agency.

12 Where data is processed

Signalment is built for veterinarians in the United States. It is hosted in the United States, in the AWS us-east-1 region, and every subprocessor in section 6 processes data in the United States. We do not currently offer data residency anywhere else.

Signalment is not offered to, and should not be used by, people in the European Economic Area, the United Kingdom or Switzerland. If you access it from outside the United States you do so on your own initiative and you are responsible for compliance with your local law, and you understand that your information will be transferred to and processed in the United States, whose privacy laws differ from your own.

13 Children

Signalment is a professional tool for licensed veterinary practitioners and is not directed to children. We do not knowingly collect personal information from anyone under 18. If you believe a child has given us personal information, email us and we will delete it.

14 Changes

We will update this policy as the product changes. When we do, the "last updated" date at the top changes and the new version is posted here.

If a change materially reduces your privacy — a new category of data, a materially new purpose, or a new recipient of your clinical content — we will email account holders at least 14 days before it takes effect, so that you can look at it and leave if you disagree. We will not apply a materially different use to content you have already submitted without telling you first.

Contact

Privacy questions, requests and complaints all go to the same address.

hello@signalment.ai

Signalment, LLC
United States