Nexora AI Privacy Policy

1. Who operates this service

Vicky Design Studio Nexora AI (“Nexora”) is a single-owner private workspace used only by Vaibhav Uttam Avachar for Vicky Design Studio’s own Cinematic Poster Design Masterclass promotions, campaigns, creative designs and analytics. It is not a customer photo-editing service or a workspace for client design projects. Only the owner operates the workspace. Publishing and automated outbound messages are currently disabled. Receiving and storing authorised inbound Instagram data is separate from permission to send messages.

Legal owner/operator: Vaibhav Uttam Avachar. Business name: Vicky Design Studio. Jurisdiction: Maharashtra, India. Public privacy contact: avacharv@gmail.com. These details were expressly confirmed by the owner for this policy. Requests should use this email; no secondary contact is designated. Effective date: 29 October 2026.

2. Information processed and purposes

Nexora processes information supplied by its owner, received through authorised platform integrations, and generated during service operation:

We use this information to provide the workspace, preserve approved originals, prepare creative/business suggestions, organise authorised enquiries, maintain security and record approvals. This policy does not grant permission to publish a design, contact a customer or use a photograph in AI processing.

3. Photos, consent and AI processing

The Hero Library contains only the owner’s own approved photographs for his appearance as the main hero in cinematic posters; it is not a customer-photo library. Uploading a hero photo stores it in the private workspace. Uploading, approving it for design selection, approving AI analysis, and approving publication are different actions. Revoking a hero's approval prevents subsequent use in guarded creative approval flows; it does not automatically erase its original, existing analysis or derived records.

Nexora currently uses Google's Gemini API unpaid text/vision service for supported analysis, chat and planning. AI requests can contain the owner's instruction, relevant campaign/context/history or memory, and, where an image-analysis request is authorised, the actual image bytes. Code requires image consent before vision submission, but it does not establish that every possible chat input is automatically stripped of personal information. The normal Instagram inbound-to-reply-draft handler uses deterministic rules; it does not send each incoming DM to Gemini. Strategy-choice requests use limited labels/counts rather than customer names or DM bodies.

Google's unpaid-service terms permit use of submitted content and generated responses to improve its products and machine learning technologies, and human reviewers may review inputs and outputs. Those terms instruct users not to submit sensitive, confidential or personal information to unpaid services. Therefore a consent checkbox alone is not a sufficient basis to send identifiable hero photos, customer messages or confidential studio data through that route. The owner has instructed that unpaid Gemini be used only for non-personal, non-confidential inputs. This restriction is an operating instruction, not a verified technical guarantee: existing image-consent paths and free-form prompts have not been shown to enforce it comprehensively. The owner must exclude identifiable hero photos, Instagram enquiries/DMs, lead details and confidential material from unpaid Gemini requests. Owning a photograph does not make an identifiable photograph non-personal. Until appropriate safeguards are separately implemented and verified, personal-data AI workflows must not be used. This policy does not assert that historical personal-data submissions have or have not occurred; no historical transfer audit has been completed. No paid service or alternate AI provider is enabled by this policy.

Nexora itself has no implemented model-training pipeline or connected, validated face-preserving raster image-generation provider. Stored correction notes, reference analyses and cached responses are application records, not proof of model training. We cannot promise that a third-party unpaid AI provider will not use submitted inputs for its own model improvement. Withdrawing consent stops future authorised processing once applied; it cannot recall submissions already processed by an external provider.

4. Pinterest references and rights

Pinterest sign-in and API read permission do not license copying creators' work, storing API content, forwarding it to AI providers, training models, or distributing derived posters. Saving another creator's Pin to the owner's board does not establish ownership of that Pin's image.

The intended long-term workflow includes studying selected Pins as visual references and potentially using references in AI-assisted poster creation. That intention must be disclosed accurately in the Developer App application. It must not be described as read-only browsing alone while enabling storage, AI transfer or derivative creation behind the scenes.

No general permission for the proposed AI-reference workflow has been established. The Pinterest integration must remain inactive for disputed storage, download, third-party transfer, AI analysis and derivative-generation uses until applicable platform permission and content rights are established and code reflects the approved limits. The existing AI-use environment flag is a technical gate, not evidence of written Pinterest permission; some reference download/storage paths are separately reachable with user consent and need review.

5. Service providers, disclosures and locations

Current infrastructure includes ChatGPT Sites/OpenAI for hosting integration and sign-in, Cloudflare Workers/D1/R2 for application execution, records and private assets, Supabase private object storage for the separate Meta edge payload/media service, Meta/Instagram for authorised platform operations, and Google Gemini when an AI request is made. Pinterest would receive OAuth/API requests only after separately approved setup and consent. The known Supabase project is in Mumbai; this does not mean all other service data stays in India. Processing may occur in other countries according to provider arrangements, which must be confirmed where relevant.

Nexora discloses only information needed for a requested service operation, such as an authorised platform API call or permitted AI request. Providers have their own privacy terms and may process technical data independently. An outbound publishing operation, if separately enabled later, would send the specifically approved creative/caption to Meta and may use a temporary signed media URL. Such operation is currently disabled.

The owner does not sell personal data, license customer photos, or share personal photographs with third parties for advertising. Nexora is not used for customer photo-editing projects. This does not exclude operational processing by the hosting, authentication, storage and AI providers described above, or an owner-authorised transfer of his approved promotional creative to Instagram if publishing is separately enabled later. Studio promotional use of the owner’s own photograph is distinct from selling photographs or providing them to advertising third parties. Only the owner operates this workspace; authorised infrastructure-provider personnel may have access under their service arrangements. No customer-photo licensing or personal-data sale function is intended.

6. Storage and security

The workspace is currently owner-private. Application queries and asset reads are scoped to the signed-in owner. Connected Meta tokens, and the implemented future Pinterest token vault, use application-level AES-GCM encryption in private object storage with separate server-held keys. Stored webhook payloads on the Meta edge also use application-level encryption. OAuth uses signed, owner-bound, expiring, single-use state; webhook processing validates signatures and account mapping. Transport uses HTTPS and private media access uses bounded signed grants where applicable.

These controls do not mean all database records or uploaded originals are separately application-encrypted or end-to-end encrypted. We do not claim that every provider log is free of personal data or that a system can be completely secure. Credential values are not intended to appear in client responses or application source. Server-side secrets and provider access require controlled administration.

7. Retention — current implementation, not an invented deletion promise

8. Access, withdrawal and deletion requests

People whose information is held may email avacharv@gmail.com, addressed to Vaibhav Uttam Avachar at Vicky Design Studio, to request access, correction, withdrawal of optional processing or deletion. We will verify the requester's authority proportionately and identify relevant active records, originals, derived files, credentials and provider-held copies. A general self-service erasure workflow is not currently available. Revoking platform access in Pinterest or Instagram stops future authorised access under that grant but does not automatically erase existing local data.

Vaibhav Uttam Avachar is responsible for reviewing requests sent to the privacy email. The proposed manual process is to acknowledge the request, verify authority using the minimum necessary information, identify matching records and related copies, review justified retention exceptions, then arrange authorised deletion or correction and verify the result. A proposed procedure is not evidence that an erasure tool already exists. We will acknowledge privacy requests within 7 calendar days and aim to complete the request or provide a reasoned status within 30 calendar days. Completion can depend on proportionate identity checks, justified retention exceptions and provider capabilities; these targets are not a promise of automatic or complete erasure. Requests involving originals require deliberate review so unrelated approved assets are not deleted accidentally. Provider-held copies and backup expiry must be handled according to verified provider capabilities; unresolved copies or exceptions must be explained to the requester. Deleting local data cannot recall a previous AI submission or automatically remove content held on Instagram/Pinterest.

9. Cookies, local storage and children

Hosting and sign-in providers may use authentication/security cookies. The inspected application service worker does not implement caching of private API responses, media, HTML or chats. We have not completed a provider-wide cookie/telemetry audit and do not claim the service is cookie-free. Browser speech recognition, if used, may involve the browser vendor; its processing arrangement must be checked before enabling personal-data voice workflows.

This is a private professional workspace operated only by its adult owner, not a children’s service. The Hero Library is limited to the owner’s own photos; it does not accept customer or minor hero photos. Google’s current Gemini API terms require adult use. Instagram enquiries can nevertheless include information from other people, potentially including minors; the owner should minimise unnecessary personal information and handle it privately. No technical age-verification control was verified, and no blanket claim is made that every Instagram sender is an adult.

10. Changes

The approved version and effective date will be published on the public policy page. Material changes in providers, data use, access, retention or AI processing require review and appropriate notice/consent before use. This policy describes the service’s data practices and current implementation limits.