1. Data we collect
- Telegram account identifiers and profile fields available to the bot, such as Telegram user id, username, first name, last name, language code, and premium flag.
- Project data: selected yard type, questionnaire answers, notes, location text or selected city, photo edit prompts, masks, uploaded photos, generated images, PDF files, and materials reports.
- Payment and entitlement data: product SKU, amount in Telegram Stars, invoice payload, masked or hashed Telegram charge identifiers, purchase state, retries, refunds, credits, and unlocks.
- Operational data: feature flags, job states, error classes, audit events, admin notes, support notes, and security logs needed to run and protect the service.
- Basic analytics and referral data, such as invite attribution, project counts, and product usage events.
Eligibility and regional compliance
Before using project creation, upload, checkout, paid, or generation flows, YardlyStudio records a minimal eligibility confirmation. The record stores user id, confirmation time, statement version, source surface, and optional trusted country decision. YardlyStudio does not store full IP addresses for this eligibility barrier.
2. Why we use data
- To authenticate Telegram users and show their projects.
- To create AI concepts, photo edit variants, PDF summaries, materials estimates, and support responses.
- To process Telegram Stars purchases, prevent duplicate payments/refunds, and recover stuck purchases.
- To prevent abuse, investigate errors, improve reliability, and maintain auditability for operator actions.
- To honor deletion, export, support, and refund requests.
3. Processors and third parties
YardlyStudio may send the minimum necessary data to processors that help provide the service:
OpenAI states that API data is not used to train or improve models by default unless the customer opts in, and that abuse monitoring logs may be retained by default for up to 30 days unless different approved data controls apply. Other providers may apply their own policies.
- Telegram / Bot Platform: bot and Mini App delivery, authentication data, messages, invoices, Stars payments, receipts, and refunds.
- AI providers: photos, masks, prompts, questionnaire text, location context, and generated outputs may be sent to OpenAI or another configured AI provider to produce results.
- Storage and hosting providers: object storage for uploads/renders/PDFs, application servers, databases, logs, and backups.
- Operational support and analytics: limited metadata needed for support, reliability, abuse prevention, and product operations.
4. Retention
- Project photos, masks, generated renders, PDFs, estimates, and questionnaire data are retained while the user keeps the project or until deletion is requested and completed.
- When deletion is completed, YardlyStudio attempts to remove private object-storage files and minimizes database records needed only for operational history.
- Payment ledger records, masked/hash charge ids, refund decisions, and audit records may be retained longer when needed for fraud prevention, dispute handling, accounting, legal obligations, and service integrity.
- Operational logs are minimized where practical and are not intended to store raw sensitive content beyond what is needed for troubleshooting, abuse prevention, or legal compliance.
5. User controls
- You may request deletion of your YardlyStudio data through the bot support flow.
- You may request an export of project and payment data that YardlyStudio can reasonably provide.
- You may stop using the bot or block it in Telegram. This stops future bot messages, but does not automatically delete retained records needed for the purposes above.
- When a policy version requires fresh consent, YardlyStudio may block upload or AI processing until you confirm the updated terms.
6. Security and minimization
YardlyStudio aims to collect only data needed to deliver the product, uses private storage for uploads and outputs, masks or hashes sensitive payment identifiers in admin views and retained records where possible, and avoids using user content to build unrelated datasets or AI products.