The Short Version
We do not sell your data, we do not share it with advertisers, we do not run ads, and we do not use advertising trackers to follow you across the web.
We collect data to run the products you choose to use, keep accounts and payments working, improve the tools, and protect the service. When product data is useful for improvement, we use privacy-preserving versions wherever practical: aggregated, anonymized, de-identified, or pseudonymized.
What We Collect
Owl + Kestrel may collect account details such as email address, display name, login sessions, communication preferences, billing and entitlement records, support messages, and security logs.
We also use first-party site analytics to count visits, pages viewed, referrers, browser locale, coarse screen size, and a pseudonymous browser identifier. We do not use advertising pixels or cross-site ad tracking.
Individual products may collect their own product data. Circadia may store sleep logs, sleep windows, wake periods, forecasts, imports, exports, and related notes. Sieve may store lists, items, votes, ratings, rankings, comparisons, preferences, and interaction events. AI products may store prompts, replies, chat transcripts, generated outputs, feedback, and debugging context.
Why We Use It
We use data to provide the products, sync or restore user state when a product supports accounts, send updates you asked for, handle payments and access, respond to support requests, debug problems, prevent abuse, and understand whether the tools are useful.
When you voluntarily share data with us for research, testing, feedback, alpha programs, or product improvement, we may use that data for research and development, including analysis, evaluation, benchmarking, and publication of aggregated or de-identified findings where appropriate.
We may use de-identified, aggregated, or pseudonymized product data to improve rankings, recommendations, forecasts, product quality, evaluations, and model behavior. We do not treat anonymization as magic; sensitive datasets still get extra care.
AI Training and Fine-Tuning
Some Owl + Kestrel products may involve AI conversations or generated content. We may use transcripts and feedback to provide the service, debug issues, evaluate quality, and improve product behavior inside that service.
Model training or fine-tuning with private chats only happens when a product clearly says so and either you have opted in or the data has been de-identified under that product's rules. Product-specific notices or settings take priority when they offer stricter controls.
Your Control and Rights
You can unsubscribe from emails at any time. Where account tools are available, you can request access, correction, export, deletion, or anonymization of personal data tied to your identity.
Some product data may be retained, deleted, anonymized, or pseudonymized according to product-specific rules. Aggregate, de-identified, pseudonymized, backup, security, accounting, or legally required records may remain after deletion when that is necessary and allowed. When that happens, we remove direct identity links where we can.
Product-Specific Terms
Some products handle more sensitive or more specialized data than the main site. Circadia sleep data, Sieve preference graphs, paid products, and AI transcript features may each have additional product-specific notices or controls.
If a product-specific policy gives you stronger protections or more specific rules, that policy controls for that product. Otherwise, this policy applies.
Technical Details
We use service providers for hosting, authentication, payments, email, storage, analytics, security, and AI infrastructure. They may process data for us only as needed to provide those services.
We use reasonable security measures for the size and stage of the project. No internet service can promise perfect security, but we try to collect deliberately, retain thoughtfully, and avoid surprise uses.
Contact
Questions, deletion requests, or privacy concerns can be sent through the contact path listed on Owl + Kestrel. If a better dedicated contact address is published later, use that instead.