JubalMed
Legal

Privacy policy

Effective November 6, 2026

In short

We collect what the product needs to work: your account, the patient records and documents you upload, the questions you ask, and a record of what the assistant did. We count page views without cookies and run no advertising trackers, our error reports carry no request contents, and we do not sell anything to anyone. Patient records are protected health information, and we handle them only under a business associate agreement with your organization.

This summary is for orientation only. The numbered sections below are the agreement.

1.Who this covers

This policy describes how Jubal, Inc. (“we”) handles information in Jubal med. Where you use Jubal med through an organization — your employer, typically — that organization controls the account, decides who may access it, and can see the work done in it. Questions about your organization’s own policies go to your administrator.

2.What we collect

Account information. Your name, email address, and the organization you belong to, together with the sign-in events needed to keep a session secure. Authentication is handled by WorkOS; we never see or store a password.

What you give the assistant. The patients you add and their records, the questions you ask, the conversations they belong to, any documents you upload, the text extracted from them, and the documents the assistant writes. This is the substance of the product and is retained so your work is there when you come back.

A record of what the assistant did. For each research run we keep which model answered, how many tokens and tool steps it took, how it ended, and any feedback you left on the answer. We also log when the assistant could not answer from the corpus, together with the question it could not answer — that record is what tells us which regulations to add next.

Technical records. Ordinary server and delivery logs (IP address, browser, timestamps) and, when something breaks, an error report. See section 3 for what those reports deliberately exclude.

3.What we deliberately do not collect

These are configuration choices in the product, not promises about future intentions:

  • No cookies for analytics and no advertising trackers. We count page views with Vercel Web Analytics, which sets no cookies and does not identify you, and we remove record identifiers from page addresses before they are counted. We do not track you across sites.
  • Error reports carry no request contents. When something fails, our error reporting is configured to exclude request and response bodies and to skip user identity entirely. A crash report tells us what broke and where; it does not carry the question you asked or the document you uploaded.
  • We do not sell personal information, share it for cross-context behavioral advertising, or hand it to data brokers.
  • Your content is not used to train models. The model providers we use serve Jubal med under zero-data- retention terms, meaning your questions and documents are not retained by them after a response and are not used to train their models. We do not train models on your content either.

4.Protected health information

Jubal med works with patient records: it reads the charts your organization uploads to prepare documents such as relatedness worksheets, certifications and IDG briefs, and it answers regulatory questions with citations. It never makes the clinical determination itself; our tier contract explains where that line sits and why.

Patient records may be uploaded only under a written business associate agreement between your organization and us. Absent that agreement, the service is not offered for protected health information, and sending it is a breach of the terms of service.

Where such an agreement is in place, its terms govern that information and take precedence over this policy to the extent they conflict. That includes restricting processing to model providers covered by an appropriate agreement.

5.Who processes it

We use the following providers to run the service. Each receives only what its function requires.

ProviderPurposeWhat it receives
WorkOSAuthentication and session managementName, email address, authentication events
ConvexApplication databaseAccount records, patients and their records' metadata, projects, conversations, documents
VercelApplication hosting and delivery, and cookie-less page-view counts (Web Analytics)Request metadata (IP address, user agent) in transit; page addresses with record identifiers removed
Amazon Web ServicesDocument storage (S3), text extraction from uploaded documents (Textract), the model gateway, and model serving (Bedrock)Uploaded documents, including patient records, and their extracted text; the text sent to and returned by models
xAIModel servingThe text sent to and returned by the model, including patient record text when a task needs it
ResendTransactional email (invitations, sign-in codes, task reminders, billing notices)Email address and message content
StripePayment processing for subscriptionsOrganization name, billing email address, invoices; card or bank account details are entered with Stripe, and we see only the type and last four digits
TypeSafe (Jev)A fast consistency check on relatedness worksheetsDe-identified worksheet rows (the HIPAA Safe Harbor identifiers are removed before anything is sent)
CloudflareDNS, and delivery of the desktop app and its updatesRequest metadata (IP address, user agent)
SentryError reportingError messages and stack traces. Configured to exclude request bodies and user identity — see “What we deliberately do not collect”.
binding.law (Jubal, Inc.)The regulatory corpus the assistant searchesSearch terms and citation lookups derived from your questions

6.Where it lives and how it is protected

Data is stored and processed in the United States. Uploaded documents are held in encrypted object storage under a key scoped to Jubal med; everything moves over TLS in transit.

Access inside the product follows your organization: a patient or a project is the sharing boundary, so people who can see it can see its records, conversations and documents, and people who cannot, cannot. Personal research conversations stay with the person who ran them.

A small number of our staff can access customer data where necessary to operate the service or to help you with a problem. That access is role-restricted and logged.

7.How long we keep it

Your content is kept while your account is active, because the point of it is to still be there. Deleted projects are recoverable for 30 days and then removed. Operational records — usage counts, error reports — are kept on shorter schedules appropriate to their purpose.

When an organization’s agreement ends, we delete or return its content on request, and delete what remains within a reasonable period afterwards except where law requires us to keep it.

We keep encrypted backups so that an outage or a mistake cannot lose your work. Content you delete persists in those backups for up to 35 days, and in a monthly copy for up to 12 months, and is then gone; backups are encrypted under keys we control, cannot be altered during their retention period, and are used only to restore service.

8.Your choices

You can see and edit your account details in the product, and delete your own conversations, memos and documents. For a copy of your information, or its deletion, contact your organization’s administrator or write to us at hello@jubal.med. Where your organization controls the account, we will generally direct the request to them.

Depending on where you live you may have rights to access, correct, delete, or port your personal information, and to complain to a regulator. We honor those rights without charging you or degrading the service for asking.

9.Changes and contact

If we change this policy in a way that materially affects how we handle your information, we will say so — a new date at the top is not notice on its own. Questions go to hello@jubal.med, or by post to Jubal, Inc., c/o Kenneth J. Ronan, Esq., Lavalle, Brown & Ronan, P.A., 750 South Dixie Highway, Boca Raton, FL 33432.