Privacy Policy

Effective Date: August 25, 2026 Last Updated: September 16, 2026

This Privacy Policy explains how Multipurpose Fun, Inc., a Delaware corporation doing business as Push Drop Stick ("Push Drop Stick," "we," "us," or "our"), collects, uses, stores, and shares information when you use the Push Drop Stick mobile application (the "App"), the companion Apple Watch app, our website at pushdropstick.com (including the signed-in web app), and the AI features we offer through them — the Coach and the Live Trainer — together with our developer API and MCP endpoints (collectively, the "Services"). By using the Services, you agree to the practices described here.

Push Drop Stick is a connected service. Your account, your pets' profiles, your training sessions, the plans you write, your conversations with our AI features, and the photos, video frames, and recordings you create inside the Services are stored on our servers so they sync across your devices and so our AI features can use them. We also use that data to improve the Services and to train our own machine-learning models. This is a change from earlier versions of the App, which kept most data on your device. The sections below explain exactly what we collect, why, who processes it, and what your choices are.

We do not sell your personal information, we do not show ads, and we do not use your information for cross-app or cross-site tracking.

---

1. Summary in Plain Language

• We keep an account for you (email, name, optional public handle).
• Your training data lives on our servers: pets, sessions and stats, goals, cues, treats, notes, and the plans you write — plus the session recordings the App makes while you train.
• AI features see what you give them. Messages, camera bursts, and camera-roll videos you share with the Coach are sent to Anthropic (Claude). When you hold the microphone button to dictate to the Coach, that audio is streamed to xAI for speech-to-text, and the Coach's spoken replies are synthesized by xAI from the reply text. During a Live Trainer session, your microphone audio and camera frames stream to GPU servers we operate on RunPod, where they are transcribed and interpreted in real time.
• We train on it. Live Trainer frames and transcripts, session recordings, conversations, and plans are used to evaluate and fine-tune our own models (for example, the small model that powers the Live Trainer). Accepting our Terms of Service includes agreeing to this use. If the law where you live gives you a right to object, Section 9 explains how.
• Some content is public by design: a plan you mark public appears on your public profile, can be copied ("forked") by other users, and is indexed by search engines and AI assistants. Photos, video clips, and recordings are stored at unlisted links — see Section 6.
• You can delete your account and data from your account settings or by contacting us (Section 7). No ads, no data brokers, no sale of personal information.

2. Information We Collect

a. Account Information

When you create an account we collect your email address, a display name (optional), and — if you claim one — a public handle for your profile page. Sign-in uses emailed magic links, one-time login codes, or a password (stored only as a salted hash). We keep short-lived login tokens, the tokens that hand you off between the App and the website, and any personal API keys you create (stored hashed) for the developer MCP server.

If you invite someone to co-train a dog, we store the email address you enter for that invitation so we can deliver it and link the account that accepts it.

b. Pets and Training Records

For each pet you add we store its name, species, breed(s), birth year, sex, and portrait photo, and which people share it. As you train we store, per pet and plan: session position and statistics (repetitions, passes and fails, success rate, rate of reinforcement, timing, streaks), goals and when they were achieved, the cue words you assign, the treats you log (including "do not use" notes such as allergies), and the notes the Coach keeps about each pet. This data is attributed to your account so it can sync across your devices and be shared with co-trainers of the same pet.

c. Session Recordings and Camera Frames

While you run a training plan, the App produces a compact, structured session recording — the sequence of actions, timings, repetition results, goal and cue names, and which reinforcer was in use. Recordings are uploaded to our storage automatically (including for sessions you abandon or that end when the App is closed) and are attributed to your account and pet. If you use the camera during a session, the App may capture short bursts of still frames and upload them alongside the recording.

d. Coach Conversations

Everything you send to the Coach — your messages, camera bursts (about ten frames), frames condensed from videos you pick from your camera roll, and any text the AI writes back, including the tools it used and their results — is stored with your account as a conversation history. The Coach may also write and revise notebook files about you and your pets that persist between conversations. Coach can perform web and scholarly searches on your behalf; the search queries it composes (which may include details from your conversation) are sent to the search providers listed in Section 5.

e. Live Trainer (Voice and Camera)

The Live Trainer is a real-time voice feature. While a live session is running, the App streams your microphone audio and periodic camera frames directly to a GPU server we operate. Audio is transcribed on that server and is not stored as audio; the text transcript of what you said, the frames the model saw, what the model said back, and a session manifest (your user ID, pet, plan, session and device identifiers, timestamps, and model/version details) are saved to our storage as a session log. Because iOS runs the microphone in full-duplex mode, audio continues to stream while the Live Trainer is speaking. Anything audible in the room — including other people — may be transcribed; see Section 9 for your responsibilities when recording others.

f. Photos and Videos You Upload

• Pet portraits you choose from your photo library.
• Demo clips you film or upload to illustrate a step in a plan you write. We trim them and strip location and device metadata from the file before publishing.
• Camera-roll videos you share with the Coach are condensed to a small set of still frames on your device; only those frames are uploaded (see 2(d)).

g. Plans You Write and Public Content

Training plans you author (directly, through the Coach, or through the developer MCP server) are stored in a version-controlled library with their full edit history. Plans have a visibility setting. A plan marked public — the default once you have claimed a handle — is listed on your public profile at pushdropstick.com/u/<handle>, can be viewed and forked by any user, is included in our public plan search, and is exposed to search engines and AI assistants through our public catalog and public MCP endpoint (which return the plan's content and your handle as its author). Plans that meet our editorial standards may also be promoted into the shared library the App offers to all users. Comments and posts you make on plans are likewise stored and shown to the audience of that plan.

h. Billing Information

Subscriptions and one-time purchases are handled by Stripe, our payment processor. Payment card details are entered on Stripe-hosted pages and never touch our servers. We store your Stripe customer ID, subscription ID, plan, status and renewal date, the email associated with your billing profile, and whether you have added a card to claim the free starter energy. Stripe may collect your card's country and other fraud signals under its own privacy policy.

i. Device, Diagnostic, and Usage Data

• Crash and performance data (Sentry). The App, the website's support page, our servers, and our GPU servers report crashes, errors, and performance traces to Sentry, including device model, OS and App version, locale, and a random install identifier. In-app product metrics (for example, counts of sessions started or plans shared) are also sent through Sentry and are not tied to your name or email.
• Push notifications. If you enable notifications we store your Expo push token, platform, and device name so the Coach can reach you with check-ins and reminders.
• Over-the-air updates. The App periodically checks for JavaScript updates from a service we host on Cloudflare; that service sees your IP address, App version, and update channel.
• Server access logs. Our servers record IP address, timestamp, resource requested, response status, and user-agent for each request, for security and capacity planning.
• AI usage ledger. For every AI turn we record the model used, token counts, and cost so we can meter your energy balance and detect abuse.

j. Support and Feedback

If you contact us by email or through the feedback form, we receive whatever you send us — including your email address, message, and any screenshot you attach — and use it to respond.

k. Information Processed Only on Your Device

The following never leave your device: on-device text-to-speech for the Live Trainer's voice (the model files are downloaded from our own object storage, which sees your IP address at download time); the on-device subject detector that crops photos around your dog; clicker and voice-command detection; your Apple Watch workout data written to Apple Health (we do not read or upload health data); and the record of which version of these terms you accepted.

3. How We Use Information

We use the information above to:

• Provide, operate, sync, and secure the Services and your account;
• Run the Coach and the Live Trainer, including giving them the context (pets, sessions, notes, frames, transcripts) they need to help you;
• Store and display your plans, recordings, and media, and share them with co-trainers and — for public content — with other users and the public;
• Meter AI usage, process payments, prevent fraud and abuse, and enforce our Terms;
• Send transactional email (sign-in links, invitations, receipts) and, if you opt in, push notifications;
• Diagnose crashes and performance problems and understand which features and plans are used;
• Evaluate, improve, and train our machine-learning models (see below);
• Respond to support requests and comply with legal obligations.

Using Your Data to Train Our Models

We build and fine-tune our own models — in particular the compact model that powers the Live Trainer — and we use content generated through the Services to do it. The training and evaluation corpus may include Live Trainer session logs (camera frames, transcripts of what you said, and what the model said back), session recordings and statistics, Coach conversations and notebook files, camera bursts and condensed video frames, and the plans you write.

• This is part of using the Services. By accepting our Terms of Service you agree that we may use your content for these purposes. There is no in-app toggle to exclude your data.
• Safeguards. Datasets are held in access-controlled storage. Before content is assembled into a training set we remove direct identifiers (such as your email address and name) where doing so does not defeat the purpose; account and pet identifiers may be retained to link a session's frames, transcript, and outcome. Raw training datasets are shared only with the compute providers who run training jobs on our behalf. We do not use your content for advertising and do not sell it.
• Trained models are derived works. A trained model does not contain your individual sessions, but its behavior is influenced by them and that influence cannot be removed once a model has been trained. We may use, distribute, and license trained models in the Services and in other products we offer.
• Your right to object. If you live in a jurisdiction that gives you the right to object to this processing (for example, the EU, EEA, or United Kingdom), or you simply prefer we not include your data in future training runs, email us at the address in Section 13. We will exclude your account from datasets assembled after we receive your request. This does not affect models already trained.

4. AI Features and Model Providers

Our AI features are powered by a combination of third-party models and models we run ourselves:

• Coach — built on Anthropic's Claude models, called through Anthropic's commercial API. Your messages, frames, pet data, and notes are sent to Anthropic to generate responses. Coach may also use smaller models to title and summarize conversations.
• Live Trainer — a Gemma-based model with our own fine-tuning, running on GPU servers we rent from RunPod and control. Your audio and frames go directly from your phone to that server.
• Research helpers — when the Coach searches the web or the scientific literature, queries go to Exa (web) and OpenAlex (scholarly).

These providers process data on our behalf under commercial terms that limit their use of your content to providing the service to us. We do not authorize any of them to use your content to train their own models. Where a provider retains data briefly for abuse monitoring, that retention is governed by the provider's terms.

AI output can be wrong, incomplete, or unsafe for your particular dog. See our Terms of Service for the responsibilities you keep when using AI features.

5. How We Share Information

We share personal information only as follows:

With service providers ("sub-processors") who process it for us and are bound to protect it:

With other users. People who share a pet with you see that pet's profile, sessions, recordings, frames, and Coach notes. Anyone can see your public handle, public profile, and public plans (and can fork them). Comments you post are visible to the plan's audience.

With the public and with machines. Public plans are exposed through our website, sitemap, public plan search, /.well-known AI catalog, and unauthenticated public MCP endpoint, and may be crawled, indexed, and reproduced by search engines and AI assistants outside our control.

For legal reasons — to comply with law, subpoena, or valid legal process, or to protect the rights, property, or safety of us, our users, or others.

In a business transfer — if we are involved in a merger, acquisition, financing, reorganization, or sale of assets, information (including trained models) may be transferred as part of that transaction, subject to this Policy.

As aggregated or de-identified data — we may create, publish, share, or sell aggregated or de-identified statistics and trained models that do not identify any individual, device, or household.

6. Public Content and Unlisted Media Links

Media files — pet portraits, camera frames, demo clips, and session recordings — are stored at unlisted URLs: long, random, unguessable links that are not indexed or listed, but that anyone who has the exact link can open without signing in. Pet portraits and demo clips are intended to be visible on public plans and profiles. Recordings and frames are only linked from within your account and your co-trainers' accounts; treat those links as private and do not share them if you do not want others to see the content.

Plans default to public once you claim a handle. You can change any plan to private in the plan's settings; doing so removes it from your public profile and from public search, but copies other users already forked remain theirs, and copies made by third-party crawlers are outside our control.

7. Data Retention and Deletion

• Account and training data (account details, pets, sessions, goals, cues, treats, plans, conversations, notebooks) are kept for as long as your account exists.
• Recordings, frames, media, and Live Trainer logs are kept for as long as your account exists, and content already incorporated into a training dataset or trained model may be retained after deletion in that form (see Section 3).
• Billing records are kept as long as needed for tax, accounting, and fraud-prevention purposes, typically seven years.
• AI usage ledger and Stripe event records are kept as operational analytics without a link to your deleted account.
• Diagnostic data (Sentry) is retained per Sentry's defaults (typically 30–90 days).
• Login tokens and codes expire automatically; server access logs are retained for a limited operational period.
• Database backups are encrypted and rotated on a short cycle; deleted data ages out of backups on that schedule.

Deleting your account. You can delete your account from your account settings in the App or the web app, or by emailing us from the address on your account (Section 13). We will delete your account and the personal data associated with it within 30 days, except data we must keep for legal, billing, or security reasons and data that has already been de-identified or incorporated into a trained model. Deleting your account removes your public profile and unpublishes your plans; forks other users made before deletion remain with those users. You can leave a shared pet or remove a co-trainer at any time in the App. Deleting the App from your device does not delete your account.

8. Security

We use administrative, technical, and physical safeguards designed to protect your information: encrypted disks on our servers, encryption in transit, hashed passwords and API keys, encrypted backups, access controls, and signed, short-lived session tokens for the Live Trainer. No system is completely secure, and we cannot guarantee absolute security. If you believe your account has been compromised, contact us immediately.

9. Your Rights and Choices

Access, correction, export, and deletion. You can view and edit most of your data in the App and the web app, and delete your account from your account settings. To request a copy of your data or correct something you cannot edit yourself, contact us (Section 13). We will verify the request using the email on your account.

Objecting to model training. See Section 3. Email us and we will exclude your account from future training datasets.

Permissions and features. Camera, microphone, photo library, and notification access are controlled by your device settings; the Coach's vision and the Live Trainer simply do not work without them. You can decline push notifications at any time.

Recording other people. Camera frames and Live Trainer audio may capture other people and their voices. You are responsible for obtaining any consent required by the laws where you are before using these features around others.

Public content. You control the visibility of each plan and whether you claim a public handle.

Residents of the EU, EEA, UK, and Switzerland. We process your data to perform our contract with you (providing the Services), on the basis of our legitimate interests (securing and improving the Services, including training our models — which you may object to as described above), to comply with legal obligations, and with your consent where we ask for it. You have the rights of access, rectification, erasure, restriction, portability, and objection, and the right to lodge a complaint with your supervisory authority. Because we are located in the United States, your data is transferred internationally as described in Section 10.

California and other U.S. state residents. We do not sell personal information or share it for cross-context behavioral advertising, and we have not done so in the preceding 12 months. You have the right to know what personal information we collect, to delete it, to correct it, and not to be discriminated against for exercising those rights. Use the contact in Section 13 to make a request; an authorized agent may act for you with written permission.

10. International Data Transfers

We are a United States company. Our servers are located in Vint Hill, Virginia, USA; our sub-processors are located primarily in the United States; and some of our team works from Canada. By using the Services you understand that your information will be transferred to, stored, and processed in the United States (and may be accessed from Canada), whose privacy laws may differ from those where you live. Where required, we rely on appropriate safeguards (such as standard contractual clauses) for such transfers.

11. Children's Privacy

The Services are for adults. We do not knowingly collect personal information from anyone under 18, and our Terms of Service require users to be at least 18 (or the age of majority where they live). If you believe a minor has created an account or provided us information, contact us and we will delete it.

12. Changes to This Policy

We may update this Privacy Policy from time to time. Non-material changes are reflected by updating the "Last Updated" date. For material changes — for example, new categories of data, new purposes, or new kinds of recipients — we will ask you to review and accept the revised Policy in the App or on the website before you continue. Your continued use after a change takes effect constitutes acceptance.

13. Contact

Multipurpose Fun, Inc., d/b/a Push Drop Stick Attn: Allison Wells [email protected]

Please put "Privacy" in the subject line for data requests.