Playbook ONE Data Usage Disclosure

Effective Date: January 1, 2026 Operated by: Blue Liquid Tech LLC Contact: hello@playbookone.app

This is a plain-language, itemized breakdown of exactly what data Playbook ONE collects, where it comes from, why, who it's shared with, and how long we keep it. It's more granular than the Privacy Policy, which it supplements - if anything here and the Privacy Policy conflict, the Privacy Policy controls for legal interpretation, and this document should be corrected to match.

1. Data Inventory

Category	Examples	Collected From	Purpose	Shared With	Retention
Account credentials	Email, hashed password, role	User at signup	Login/authentication	Supabase (processor)	Life of account
Player/roster profile	First/last name, email, gender, position, jersey number, team/league/season	Player, parent/guardian, or league admin/owner	Roster management, display, scoring rules	Supabase; visible to teammates, league staff	Life of account/league history
Game events & stats	Play-by-play events, per-game/season/career stats, performance scores	Scorekeeper entry (manual or voice-assisted)	Core product function - stat tracking	Supabase; visible per league's sharing settings	Retained indefinitely as league/career history unless deletion requested
Payment/transaction records	Amount, date, status, payment reference ID (not full card number)	Stripe, at checkout	Season/registration billing, support, accounting	Stripe (processor); Blue Liquid Tech retains transaction metadata only	Per accounting/tax requirements
Game video	Uploaded footage, may include player likeness	League/team upload	Game review, highlights	Cloudflare Stream (processor)	Until league/uploader deletes it
Voice transcript & AI parse	Transcribed play text, Claude's structured interpretation, confidence score, scorekeeper's confirm/edit/reject action	Scorekeeper App voice input	Fast live stat entry; audit trail for disputed plays	Anthropic (processor, receives transcript text only - not audio); Supabase (stores the audit log)	Tied to the game record
Communications	Emails sent to you, correspondence you send us	System-generated or user-initiated	Account/league notices, support	Our email delivery infrastructure	As needed for support history
Device/usage/log data	Device type, OS, app version, crash logs	Automatically, via app use	Debugging, reliability	Railway (hosting), Expo/EAS (build/update delivery)	Standard log retention of underlying providers

2. What We Explicitly Do Not Collect

Raw voice audio recordings (speech-to-text happens on-device; only the resulting text is sent onward - see Section 3)
Biometric identifiers: no facial recognition, no voiceprints
Government ID numbers, Social Security numbers
Precise/continuous geolocation
Home address or phone number as part of standard registration
Health or medical information

3. How the Voice-Assisted Scoring Feature Actually Works

Because this is the most novel - and most AI-adjacent - data flow in the product, here's the step-by-step:

A scorekeeper holds a button in the Scorekeeper App and speaks a play (e.g., "12-yard reception, Jordan Lee").
The device's own built-in speech recognition (via expo-speech-recognition) converts that speech to text locally on the device.
Only the resulting text string is sent to our backend (/parse-play), which forwards it to Anthropic's Claude Haiku model to parse into a structured play (event type, player, yardage, etc.).
Claude's structured suggestion is returned to the scorekeeper's screen for review.
The scorekeeper must confirm, edit, or reject the suggestion - nothing is written to the official game record automatically.
Whatever the scorekeeper actually approves is saved via /confirm-play, and the full round-trip (original transcript, what Claude proposed, what the scorekeeper approved) is written to an audit-log table (voice_play_audit_log) tied to that specific game and scorekeeper.

We do not use this transcript data to train AI models, ours or Anthropic's - it's used solely to parse that one play in that one moment, per Anthropic's standard commercial API terms (API inputs are not used for model training absent separate opt-in).

4. Subprocessor List

Subprocessor	Function	Data Category Involved	Location
Supabase	Database, authentication, file storage, realtime sync	Nearly all categories above	Hosted infrastructure (US region)
Stripe	Payment processing	Payment/transaction data	US
Cloudflare	Video streaming/CDN	Uploaded game video	Global CDN, US-operated
Anthropic	AI parsing of voice-transcribed play text	Voice transcript text only	US
Railway	Application hosting	All in-transit/at-rest data for running apps	US
Expo / EAS	Mobile app build & update delivery	Device/app diagnostic data	US
GoDaddy	Domain/DNS management	None (infrastructure only, no user data)	US

We'll update this list if we add or remove a subprocessor that handles personal information, and material changes will be reflected in the Privacy Policy's "Changes" section.

5. Data Access Within a League

Playbook ONE is a shared platform within a league - this section clarifies who can see what:

Players see their own full profile and stats, and typically their teammates' names/stats/roster info as part of normal team use.
Scorekeepers see roster and live-game data for the games they're assigned to score.
League Admins see roster, stat, and game data for their league.
League Owners see everything their admins see, plus registration/payment status for their league (not full payment card details - those stay with Stripe).
Some leagues opt into a public schedule page (no login required) that shows schedule and, depending on league settings, limited public stats - this is controlled by the league owner, not on by default for personal player data beyond what's needed for a public schedule.

6. Your Choices

See Privacy Policy Section 6 for how to correct, access, or request deletion of your data, and Section 5 there for children's-privacy-specific rights.

Contact: hello@playbookone.app Blue Liquid Tech LLC - governed by the laws of the State of Illinois.