Privacy

Your answers and notes in the H.U.M.A.N. reflection are held only in this browser tab's memory. They are not written to browser storage, cookies or the page address, and they are not sent to our server. Closing or refreshing the tab clears unsaved answers.

You can choose "Save progress to a file" to download your answers to your own device. Loading that file later reads it on your device only; it is not uploaded. Keep the file private, since anyone with it can read your answers.

Results PDFs, print views and the private history comparison are all created in your browser.

Inside the plan step you can ask for an AI-drafted development plan. Nothing is sent unless you press the button and confirm. Before sending, the page shows exactly what will go: your typed reflection notes (up to 2,000 characters), your chosen development stage and the published suggested actions for your priorities. Assessment answers, scores, context details and other notes are never included.

The request goes through Lovable AI to an OpenAI model. Those providers process the text under their service terms; we cannot promise they keep no logs. The draft is shown only on your screen and is not saved. You can always build your plan without AI.

The Kingsmen Lab has a separate optional coaching feature with its own preview and confirmation each time.

The Kingsmen Lab saves your reflections on your own device by default, with a clear "Saved on this device" indicator. You can choose not to save, export your data, or delete everything. See the Kingsmen privacy page.

The Educator Leadership Lab (opens in a new tab) is a separate app with its own storage behavior. Check its own notice for details.

This site counts visits and pages viewed through its hosting provider. It also uses PostHog to record a short list of actions: clicks on certain buttons, when a reflection starts and finishes, opening a lab or the facilitator guide, sharing a public link, downloading files, and whether an update signup succeeded or was rejected.

Entering the Kingsmen Lab also records its public page path and the referring website address if your browser supplies one. We remove referral paths, extra parameters and fragments before sending it.

Each record contains only the action name, a short label for the button or place, the question set used, the page address without any extra parameters, and an anonymous id that lasts only while the tab is open. No cookies are set for this. Answers, scores, notes and emails are never part of these records. Your browser still sends normal technical details, such as its IP address, to the analytics provider when a record is sent.

The footer of each page offers a voluntary signup for occasional updates about new guides, tools and site news. It collects only your email address, which is stored securely in this project's backend and used only for those updates.

Your signup is never connected to your reflection answers, scores, notes or lab work, and we never sell or share the list. Every update email includes an unsubscribe link, and you can remove yourself anytime on the unsubscribe page. Signup attempts are counted anonymously by outcome only (for example, succeeded or rejected as spam); the email address is never part of those counts.

The Conversation Lab holds drafts, branch selections, context, participant transcripts, plans and self-reviews only in this session’s memory. Refresh, closure or leaving the Lab clears unsaved work. You can explicitly save a validated JSON text file, download your plan or transcript, or print. Files you download stay on your device after reset. Diagnostic and Kingsmen files remain separate.

Optional recording captures your microphone only after you start and grant permission. Audio remains in browser memory for local playback and explicit download, never in browser storage or on our server. Recording does not automatically transcribe or send audio. The latest two takes per turn are held until deletion, reset or leaving the Lab; temporary playback links are released. Transcripts are participant-provided, editable and not verified against audio.

Optional dictation detects your browser's speech recognition support. It starts only after your informed choice and microphone permission, offers stop/discard and requires confirming an editable transcript before adding it. Your browser or platform vendor may process audio on its servers under its terms. The Lab does not receive that audio, cannot verify the vendor's retention period and does not claim on-device processing. Typing and device-keyboard dictation remain alternatives.

Optional text-only AI review has its own editable approved transcript, exact request preview and separate consent. The route is this site's server, Lovable AI Responses, then OpenAI GPT-6 Astra with store: false. That option disables stored Responses application state, not all service logs. OpenAI's API data controls describe no training by default and abuse-monitoring retention up to 30 days, subject to documented exceptions and approved controls. The workspace catalog labels this endpoint zero-retention, but the Lab cannot independently verify the contractual controls or Lovable gateway payload-log retention. Only approved text and fictional context are sent, never audio. Cloud stores only opaque quota keys, counts, windows and temporary leases for abuse prevention, not participant content or raw IP addresses. An essential signed 24-hour cookie limits reviews, never analytics. Hour/day windows expire within 2/48 hours and leases within 20 minutes; expired records are removed on the next admission. Limits are 3 per hour, 6 per day per session, 40 globally per UTC day and 2 simultaneous reviews. No participant accounts are added.

Official sources: OpenAI API data controls (new tab), Lovable AI (new tab), and MDN speech recognition (new tab). The feature disclosures link browser-vendor notices too.

No Conversation Lab actions, text, context, branches, transcripts, audio or reviews are sent to PostHog; the application has no PostHog replay or autocapture SDK. Hosting still receives ordinary page requests and technical details. Sensitive content is kept out of addresses, metadata and routine error messages.

Research editor sign-in is separate from visitor reflections. Authorized editors can store public-source citation metadata, verification checks, notes and AI explanation drafts. Only approved source records and approved explanations are visible publicly. There are no participant accounts or profile fields in this workflow.

Optional research questions and replies stay in this tab's memory, not this site's database, browser storage or analytics. Each question requires consent before it is sent with verified source evidence and methodology notes to Lovable AI and OpenAI Responses with store: false. That setting does not eliminate every provider or gateway service log. Do not include identifying details or private assessment answers. The assistant does not read assessment or lab state.

A separate essential signed 24-hour cookie and opaque quota counters limit questions to 3 per hour and 6 per day per visitor. Research AI shares a 40-request UTC-day cap and 2 simultaneous requests across editor and visitor uses. Expired counters are eligible for bounded cleanup on later requests; hour/day records expire within 2/48 hours and leases within 20 minutes. No question, reply, raw IP address or referral URL is stored in these counters. Optional speech uses browser recognition with separate consent, not audio uploads.