Privacy policy

Effective 9 September 2026

1. Who is responsible

Jett is operated by Sandro Gantze as an individual. No incorporated company is identified as the operator. For privacy requests, contact sandro@jett-mail.com or support@jett-mail.com.

This notice covers the Jett website, web and desktop email clients, including information about people who correspond with a Jett user. GDPR and DSGVO refer to the same EU data protection regulation.

Jett determines the purposes of account administration, billing, security and website analytics. Where an organisation uses Jett to process workplace correspondence on its instructions, that organisation may be the controller and Jett a processor. Business use requires the appropriate data processing agreement and an assessment of those roles; these terms alone are not that agreement.

2. Information and sources

• Account and connection data: name, email address, profile image, account/session identifiers, connected mailbox details, access tokens and preferences, from you, Clerk and your connected provider through Aurinko.
• Mailbox data: sender/recipient addresses, subjects, timestamps, flags, headers, short body snippets, attachment names and other attachment metadata, labels, priorities, summaries and suggested actions. Correspondence can contain information about people who do not use Jett.
• Content: messages and attachments you open or use, text you submit to AI, and locally saved drafts, chat history, calendar entries, Library items, Reader edits and Personal Context where available.
• Billing and support: Stripe customer/subscription identifiers, plan/payment status, membership eligibility where offered, and information you send in a support request. Stripe collects payment details through its payment interface.
• Website and operational data: network information such as IP address and user agent, request times, error/performance diagnostics and usage/cost records. With analytics consent, PostHog receives website interaction data and limited product events. Waitlist email addresses are submitted to Appwrite.

3. Local storage and cloud processing

Your connected email provider remains the source of your messages. Jett accesses it through Aurinko. Opening mail or attachments may pass their content through Jett’s authenticated server and Aurinko before it reaches your device.

Jett caches message bodies and attachment files on your device. Desktop uses SQLite; supported browsers use SQLite in browser storage, with IndexedDB as a fallback. Local search and mailbox analytics use this device data. Offline availability depends on what has been cached, your session and your device’s storage.

Jett’s Supabase/PostgreSQL database stores mailbox metadata, including subjects, snippets and AI summaries, alongside account, synchronisation, billing and job records. These records can reveal personal or confidential information. Full raw email bodies and attachment files are not stored in that database.

Local-first does not mean all processing stays on your device. Jett does not provide end-to-end encryption between email correspondents, and does not promise application-level encryption of every local database. Protect your device account and disk, especially on shared computers.

4. How AI uses information

• Automatic sorting and summaries: when background classification is enabled for your account, a worker retrieves message content and sends sender/recipient information, subject, bounded message text and relevant label instructions to Microsoft Foundry. This can run without you opening each message. The cloud retains resulting labels, priorities, summaries and suggested actions; temporary supporting excerpts are discarded by the classifier.
• Text chat: the basic Quick Agent sends your prompt and a limited recent conversation to Microsoft Foundry. It has no automatic mailbox search or attachment access. Completed chat history is stored locally.
• Reader analysis: an explicit analysis request sends extracted, bounded document page text to the model provider. The AI request does not upload the PDF file itself. Extracted text can still contain sensitive information.
• Preview drafting and Personal Context: mailbox-aware drafting and Reply Review remain subject to release restrictions. Where available, separately authorised selected-mail text and accepted writing preferences may be sent for a draft. Reply Review is opt-in and can request drafts in the background while the device is active. Personal Context learning is off by default; its analysis uses locally cached sent mail and suggestions require review. It is not training a shared foundation model.

The current model route uses Microsoft Foundry. Semantic embeddings and alternative providers are not described here as active services. Changing these processing purposes or recipients requires an updated notice and any required authorisation before activation.

Jett does not use mailbox content to train a general-purpose AI model or sell it for advertising. Microsoft describes training restrictions for models sold by Azure, but provider abuse monitoring can include content retention and authorised human review. Exact deployment terms matter, including for previews. A stateless Jett request does not guarantee zero retention by the provider. See Microsoft’s model data-processing explanation.

Labels and drafts are suggestions and can be wrong. Jett’s AI does not make decisions intended to have legal or similarly significant effects on you. Draft generation does not send an email. Sending and other supported external actions require the applicable user action or confirmation; ordinary background classification can apply labels.

5. Purposes and legal bases

• Providing the service you request: account access, mailbox connection, synchronisation, requested features, support and subscription administration rely on Article 6(1)(b) GDPR where objectively necessary to perform our contract with you.
• Security and reliability: preventing abuse, diagnosing failures and protecting accounts may rely on Article 6(1)(f), subject to a balancing assessment. The interests are a secure, reliable service and defending legal claims.
• Legal duties: required accounting records and legally required disclosures rely on Article 6(1)(c).
• Optional analytics and communications: analytics consent and any separately requested marketing consent rely on Article 6(1)(a). A waitlist request is limited to communications about that request, not unrelated marketing.

Accepting terms or connecting a mailbox is not blanket consent for every AI use. Optional AI processing, correspondence about other people and sensitive information require a purpose-specific legal assessment. A user contract alone does not supply a legal basis for every correspondent’s data or satisfy Article 9 for health and other special-category information. Jett limits these uses to the features described here and applies additional safeguards where required by law.

Account and connection information is necessary to use connected mail. Optional analytics is not required to use Jett. You can revoke mailbox access in your provider’s account settings; this stops future access but does not itself erase existing copies or cancel a subscription.

6. Service providers and international access

• Clerk: sign-in, account identity and sessions.
• Aurinko and your email provider (such as Google or Microsoft): mailbox authorisation, synchronisation, message/attachment retrieval and requested mail operations.
• Supabase: database and synchronisation infrastructure; Vercel: website/application hosting and server requests.
• Trigger.dev: background worker execution. Job payloads use references; running workers may transiently process mail text for classification.
• Microsoft Foundry: the AI processing described above.
• Stripe: payment processing, billing and subscription management.
• PostHog: consent-based analytics; Sentry: application error/performance monitoring where configured; Appwrite: waitlist storage.

Recipients can also include people you email, publishers contacted by a confirmed unsubscribe action, and authorities where legally required. Opening third-party sites, feedback services or remote email images can disclose network information to their operators under their own policies.

Providers may process data or allow support access outside the European Economic Area, including in the United States. EU infrastructure placement alone does not guarantee EU-only processing. Foundry Global deployments, for example, can process prompts across geographies.

The transfer mechanism depends on the provider and recipient. It may be an applicable adequacy decision or safeguards such as the European Commission’s standard contractual clauses. Contact the operator for information about the mechanism applicable to a particular provider and for a copy of the relevant safeguards where available.

7. Cookies, device storage and choices

Authentication, security, your cookie choice and local storage needed for requested features use cookies or device storage. Where strictly necessary, the device-access exemption is § 25(2) TDDDG; the GDPR legal basis depends on the purpose above.

PostHog loads only after “Allow analytics”. It measures website use and limited product feature events. Session recording is disabled, page URL queries are removed from captured events, and mailbox text is not an intended analytics field. Network requests still expose connection information to the recipient.

Choose “Reject analytics” to decline, or open “Cookie settings” in the website footer to change or withdraw your choice at any time. Withdrawal stops future collection without affecting the lawfulness of earlier processing. The choice is stored in this browser for up to 180 days. Other browsers and devices have separate choices.

Sentry’s application monitoring is separate from optional PostHog analytics. Its configuration strips user details and request bodies and disables AI input/output capture. Diagnostic events can still contain technical error information. Jett uses those events to investigate faults, protect the service and improve reliability.

8. Retention and deletion

Local caches may be evicted to enforce storage limits or removed through device/browser storage controls. Clearing local data can remove unsent drafts, local chat history, Reader edits and preferences. Signing out or uninstalling should not be assumed to erase every local copy, export or backup.

Cloud mailbox records support the connected account and synchronisation. Account-wide deletion is handled by request to the operator. It may take time to remove data from restricted backups and records that must be kept by law. Deleting a Jett account does not delete the source mailbox held by Google or Microsoft.

Retention must end when its purpose ends unless a legal obligation or justified claim requires limited continued storage. Billing, support, membership, diagnostics, provider logs and backups are kept only for their applicable purpose and legal retention period, then deleted or anonymised. Previously collected analytics can also be included in an erasure request.

9. Your GDPR/DSGVO rights

Subject to applicable conditions, you can request access, correction, erasure, restriction and portability, withdraw consent, and object to processing based on legitimate interests. You can object to direct marketing at any time. Contact the operator above; do not send your password or full mailbox as proof of identity.

Requests must normally be answered within one month. A necessary extension of up to two further months must be explained within that first month. Proportionate identity checks may be necessary. Rights requests are normally free.

You can complain to a supervisory authority, particularly where you live, work or where an alleged infringement occurred, without first contacting Jett. Find yours in the EDPB authority directory. See also the EDPB explanation of individual rights.

10. Age, Google data and updates

Jett is not intended for children under 16. Users under 18 need the legal capacity or guardian authorisation required for their use and purchases.

Use and transfer of information received from Google APIs must follow the Google API Services User Data Policy, including Limited Use requirements. Mailbox access is for user-facing email features, not advertising or general model training.

We will date revisions and explain material processing changes before they take effect where required. A revised notice or continued use does not replace consent where required.