Practice is personal.
So is your data.

What leaves your browser

Your replies and custom scene descriptions are sent to our server to run the simulation. Don’t include real names or identifying details.

Configured text AI providers: OpenAI. Replies and recent scene context may be sent to the primary provider and, if needed, the fallback provider for tagging, child dialogue and optional scene generation or coaching. Their own data policies also apply. Configuration does not guarantee availability.

A failed model request may already have sent your text to the provider before rules and written dialogue take over.

Server sessions include replies and expire six hours after creation; further replies do not extend that window. Generated custom scenes expire after 24 hours. Daily standings store a random session ID and a score, with a 14-day expiry window. They do not store your name or reply text. The deployment’s current storage mode is available at the health endpoint.

The server uses Redis when available and falls back to process memory. Memory data can disappear on a restart and is not shared between web processes. Expired entries are unavailable immediately; memory cleanup runs every minute.

Abuse prevention uses a hash of your IP address, with limits lasting one hour or 24 hours for custom scenes. Raw IP addresses are not kept in that store. Aggregate usage counters have a 30-day expiry window and contain counts, not reply text.

What stays here

History, your playbook, streak information and preferences live in your browser’s local storage under doover:v1. They remain until you clear them. Private browsing or blocked storage may keep them only for this visit.

If you switch the home page between light and dark, that choice is kept in one cookie, doover_theme, holding only light or dark, so the page opens in the same theme next time. It is not used for tracking.

A shared playbook link carries the saved lines in its URL fragment. This application does not send that fragment to the server. Anyone who receives the link can read the lines, and copies cannot be recalled.

Clearing browser data does not delete server sessions, shared copies or waitlist submissions. It does remove the season-pass cookie.

Optional voice

Voice is part of the season pass. Recording requires your microphone permission and a supported browser. The three-second volume calibration stays in your browser. A reply recording is sent through our server to OpenAI for transcription when transcription is configured. This app does not save the recording on the server; its transcript is handled like a typed reply. OpenAI’s data policies also apply.

OpenAI transcription is configured here; availability depends on the provider. OpenAI speech playback is configured here.

For spoken child replies, the server sends the child’s line to OpenAI when speech playback is configured. Generated audio is cached in process memory. Entries expire after up to one hour or when the session expires, whichever comes first. Later playback requests clear expired entries; a server restart clears the cache. Browser playback uses your browser’s speech service, whose data handling depends on the browser and voice.

Season pass and payments

If you open the season pass, the server sets one cookie, doover_pro. It holds only how the pass was opened (demo code or test checkout), when, and when it ends after 30 days, signed so it cannot be edited. It is httpOnly, so page scripts cannot read it, and it is not used for tracking. “Return to Free” on the pricing page deletes it.

Checkout is handled by Stripe on Stripe's own page, in test mode: no real charge is possible, and no card data reaches our server. When you come back, our server asks Stripe only whether that checkout was completed.

Before the season pass existed, the pricing page offered a waitlist. Those entries (email, chosen plan, time) have no automatic expiry and persist in Redis when available.