Privacy Policy

Sahara is built for a single account holder to store and revisit personal memories (text, optional files such as images, audio, and PDFs, and optional chat imports), in-app chat (including optional “Presence” style conversation features), and related metadata. Your content is tied to your account and is not used to sell ads. We do not use your content to train public or general-purpose machine-learning models for unrelated third parties. Some features send portions of your saved content to Google’s AI services to power search, answers, cards, imports, attachments, and background enrichment, as described below. If we ever introduce optional use of your data to improve our own models, we will describe it here and, where required, obtain appropriate consent.

The data controller for the service is the person or organization operating the Sahara deployment you use (for example, the company named on the site or in your agreement). If you are unsure who that is, contact us from the address you signed up with.

Account and profile. We use Clerk for authentication. Clerk may process your email address, name, profile image, and security-related data according to Clerk’s privacy policy. We store a corresponding user record in our database (for example, email and display name) so the product can function.

Waitlist. If you join our waitlist on the marketing site, we collect and store the email address you submit so we can notify you about access. You can ask us to remove it anytime via contact.

Memories and related content. You may add text memories, optional attachments (for example images, audio, and PDFs), and people profiles (such as name and relationship). If you import a WhatsApp export, we process the file you upload to create memory entries and structured conversation data derived from that chat.

Chat, Presence, and feedback. When you use Ask, chat threads, or optional Presence-related features, we store your messages, thread metadata (such as titles and modes), and related product data (for example verification metadata, citations, or feedback you submit) so the feature can function and improve reliability.

Technical and usage data. Like most web services, our servers and hosting providers may log data such as IP address, device and browser type, timestamps, and request metadata for security, reliability, and debugging. Our web app may use Vercel Analytics to collect aggregated usage signals (as described in Vercel’s documentation).

Background jobs. Long-running tasks (such as imports, embeddings, attachment processing, or enrichment) may briefly pass payloads through a job queue (for example Redis) on infrastructure we operate.

• Provide, maintain, and secure the service (including sign-in and syncing your data).
• Store and display your memories, timeline, imports, attachments, chat threads, and person profiles exactly as you expect when using the product.
• Generate embeddings (numeric representations) of text derived from your memories, imported chats, and related content using Google’s Gemini embedding models so we can retrieve relevant material when you use features such as Ask or Presence.
• Send relevant excerpts (including memory text, derived chat lines where applicable, attachment-derived text where applicable) and your prompts to Google Gemini (including synchronous and batch APIs where we use them) to produce on-demand answers, summaries, memory cards, tagging, and other in-product outputs. This processing happens over secure APIs; we do not control Google’s subprocessors beyond their published terms.
• Comply with law, enforce our terms, and protect users and the service.

When you use Ask, Presence, memory cards, imports, attachment processing, or similar features, the service may select excerpts from memories, stored chat-derived content, attachments (for example extracted text from a PDF), and related context, together with your prompt, and send them to Google’s generative AI APIs (including real-time and, for some maintenance tasks, batch or asynchronous processing). Outputs are produced for your account and are not shared with other Sahara users.

Certain developer-oriented exports (for example structured exports related to Presence) are generated only when you request them while signed in; they contain material derived from your saved archive. We do not currently use those exports to train our own machine-learning models; if that changes, we will update this policy and provide any notice or consent required by law.

Do not submit information you are not allowed to share, or highly sensitive data you are uncomfortable sending to an AI provider, if those features are enabled on your account.

Application data is held in a PostgreSQL database operated as part of our infrastructure. Attachments and uploaded import files are stored in S3-compatible object storage (for example, Cloudflare R2 or AWS S3, depending on configuration). Providers may be located in regions you or we select.

We do not sell your personal information. We share data only as needed to run Sahara:

• Clerk — authentication and session management.
• Google (Gemini) — embeddings and generative responses for in-product features, as described above.
• Vercel — site hosting and, where enabled, Vercel Analytics for aggregated product usage signals.
• Hosting and infrastructure — servers, databases, object storage, job queues, logging, and email or monitoring tools used to operate and protect the service.
• Legal and safety — when required by law, or to protect rights, safety, and integrity.

We keep your data while your account is active and as needed to provide the service. You may delete individual memories or people within the product where those controls exist. In Settings, where available, you may download a machine-readable export of your data and delete your entire Sahara account; deletion should result in removal or irreversible disassociation of personal data according to our technical and legal obligations. You can also manage sign-in and some profile data through Clerk. Some backups may persist for a limited period before rotation.

We use industry-standard practices such as encryption in transit (HTTPS), access controls, and separation of customer data in the application layer. No method of transmission or storage is completely secure; if you discover a vulnerability, please tell us.

Depending on where you live, you may have rights to access, correct, delete, export, or restrict processing of your personal data, and to object to certain processing. You can request a data export or account deletion in Settings where those controls are enabled, or contact us with the email address associated with your account. You may also have the right to complain to a supervisory authority.

Sahara is not directed at children under 13 (or the minimum age required in your country). We do not knowingly collect personal information from children. If you believe we have, contact us and we will take appropriate steps.

If you access the service from outside the country where our servers are located, your information may be transferred across borders. We rely on appropriate safeguards where required (such as standard contractual clauses offered by our vendors).

We may update this policy from time to time. We will post the new version on this page and update the “Last updated” date. For material changes, we may provide additional notice (for example, by email or in-product message).

Questions about privacy or this policy? Visit our contact page.

You can also read our Roadmap for current feature status and visit About for the product principles behind these privacy choices.