Loading…
Loading…
Draft for the implemented beta. Operator details and provider retention schedules require completion and founder/counsel review before public launch.
Persir lets adults build, inspect, correct, deactivate, delete, and export a private profile. After a separate analysis acknowledgment, a personal ChatGPT or Claude export can supply recent visible conversation text for private, evidence-backed facts and insights. Imports are snapshots; additional information requires another import. Optional Google Analytics measures visits and feature use only after separate permission. There is no merchant access or current rewards offering.
We process your account email and authentication information, minimal Google identity information if you choose Google sign-in, profile facts you voluntarily enter, retained ChatGPT and Claude conversation text and available source identifiers, titles, timestamps and branch relationships, and necessary technical/security information. A separately acknowledged recent-history import defaults to supported messages from the last 30 days; size/count limits and exclusions can reduce that coverage. ZIPs are scanned locally with bounded browser memory. Only the normalized text dataset is transferred; the original archive, images, audio, PDFs, other attachments, hidden reasoning, tool/connector payloads and unrelated account metadata are not uploaded or analyzed. Excluded and undated messages do not become invented history. Google sign-in is authentication only; we do not request Gmail, Drive, or contacts access. Do not include private credentials.
We use this information to operate your account, save and display your profile and source records, apply your requested controls and accessibility settings, provide verification and recovery, and protect the service. After the versioned private-analysis acknowledgment, configured server workers extract explicit user statements, verify source evidence, calculate conversation observations and optionally explain supported patterns. Imported sources, facts, evidence, insights and snapshots remain private and separate from manually entered registration answers; extraction never silently overwrites those answers. Assistant text is context, not independent evidence about you; quotations, third-party material, hypotheticals and ambiguous branches are excluded from confirmed facts. We do not infer medical diagnoses or sensitive traits from indirect signals. Explicit sensitive information within acknowledged text is classified as private-sensitive. You can correct or dismiss derived information. AI can make mistakes; evidence and coverage are available for inspection. The acknowledgment creates no sale, sharing or marketing permission.
Supabase provides database, authentication, durable processing and Edge Function infrastructure. Vercel provides application hosting when the operator deploys there. Separately acknowledged private extraction sends bounded retained conversation text to the configured OpenAI API for fact extraction and evidence-support verification; no complete ZIP or attachment is sent to OpenAI. Requests use store:false and no browsing or external tools. This disables requested Responses storage, but does not establish Zero Data Retention. OpenAI states that API data is not used for training by default unless the API customer opts in; its normal abuse-monitoring logs can contain input/output and be retained for up to 30 days, with legal or safety exceptions. The operator must verify actual account retention/data-sharing settings before processing real imports. Source removal cannot retract provider processing already performed. City searches send only the typed city query to Open-Meteo; selected locations may include a time zone, with the browser zone as a fallback. This beta does not sell profile information or disclose imported/derived conversation data to merchants. Future sharing of other eligible information would require separate, scoped permission.
About Me uses supported saved profile and game evidence to create five private experiences. Missing evidence stays locked, and generated outputs and viewed states are saved per account. The life map sends only supported city names to Open-Meteo for geocoding, never conversation text, event descriptions or user identifiers. Opening the map requests visible map tiles from OpenStreetMap, which receives your network information and the requested tile coordinates. Sharing or downloading is your explicit action and includes the selected output only. These experiences create no merchant-sharing permission.
Choose whether to acknowledge private text analysis and upload an export, cancel processing, inspect/correct/dismiss derived facts in each source folder, or remove ChatGPT or Claude data independently in Data. Local preparation and transfer require the page to stay open; after verified transfer, durable server processing and temporary retries can continue when you leave. Existing source-only imports are not automatically authorized for new analysis. Inspect and edit your manually entered profile, deactivate individual answers, delete values, download the available profile JSON, or delete your account in Settings. Keep the original source export as your source copy. Deactivation retains manual values but excludes them from usable projections. Deletion removes live values. Active, imported, Available for sale and Option to sell never mean shared. Registration and conversation records remain owner-only; any future disclosure requires separate rights review and authorization.
The Data page labels individual profile items: a money icon means Potentially eligible for future sale; Personalization only means not eligible for sale through this product; Under review means eligibility has not been verified. Labels depend on applicable rights, source restrictions, provenance, and other conditions. They never mean a sale occurred or permission was granted. Registration and retained conversation records are organized under Available for sale in the owner-only Data explorer. This is an internal classification, not a listing, eligibility approval or sharing grant. Rights, provenance, sensitivity and source restrictions remain separate; no company receives these records automatically. Sales, deals, and rewards are coming soon, without guaranteed offers, earnings, or availability. Any future sale or sharing of eligible information would require a separate, explicit decision identifying the information and recipient. Account creation, imports, private-analysis acknowledgment, general Terms acceptance and a Yes interest answer do not supply that decision. Existing permissions are not broadened. About Me displays five private experiences using supported saved evidence; the separately acknowledged conversation import pipeline exposes its private source results through Data. These outputs grant no additional permission.
Daily Draft stores your private draft, original shuffled order, and final per-item tier or unfamiliar response with the dated question and factual option context. Final boards cannot be edited through gameplay. Other players see aggregate rating counts, averages, and tier distributions, excluding unfamiliar answers; they cannot read your private entry. Your submitted choices are explicit preference evidence, separate from inferred patterns, and appear in your profile JSON export. There is no personal AI analysis of these answers yet. A shared image contains the saved board you choose to share; its challenge link contains only the date. Drafts and final answers remain until account deletion, which also removes their contribution to live aggregates. No merchant-sharing permission or marketing consent is created.
If you choose to answer the optional rewards-interest survey, we store one Yes or No per account, the question version, and a database-generated timestamp to understand product interest. It does not authorize sharing, sale, or marketing and does not change permissions. The original answer is retained across sessions and cannot be changed or removed through ordinary requests to vote again. It remains until account deletion, which removes the live response along with other app-owned data. Provider backup and security-log retention follows the limits described below.
Live profile and conversation source/derived records remain until you remove them or delete your account. Removing one source deletes its live conversations/messages, derived support, insights/snapshots, extraction caches, tasks and relevant notifications, and advances that source generation so in-flight workers cannot recreate deleted data. Facts supported by another remaining source retain that support, along with your corrections and dismissals. Evidence-free correction and dismissal records may remain to honor your choices on later imports; they cannot support insights without current evidence. Minimal source-import fingerprints, counts, confirmation/disclosure versions and canceled/deleted job markers may remain until account deletion to reject stale requests; removed markers contain no retained conversation text or selected identities. Account deletion cascades through app-owned records. We keep minimal non-value metadata for manual profile changes. Provider backups and technical/security logs may remain after live deletion; actual Supabase/Vercel backup/log schedules must be configured and disclosed by the operator before public launch. OpenAI processing follows its configured account policy and the standard abuse-monitoring limit described above; store:false alone does not remove that retention.
If you allow analytics, Google Analytics 4 receives sanitized page paths, static feature identifiers and product-action events, including the Yes/No rewards-interest choice. We exclude names, emails, answers, conversation content, filenames, authentication tokens and personal insight content. GA uses its normal browser cookies and technical information to estimate visits, engagement and return visits. Advertising signals and personalization are disabled. Permission is optional and separate from any future sharing or sale decision. Declining sends no analytics events. Change this browser’s choice through Analytics preferences in the footer or Settings; withdrawing permission disables collection and clears app-created GA cookies. It does not erase earlier reports; contact the operator about prior data. Different browsers keep separate choices and identifiers. Google processes analytics data under its terms; before launch the operator must configure and disclose the property’s retention settings. No session replay is installed.
We use authentication cookies, validated user checks, database access policies, and restricted server credentials. No system is absolutely secure. Optional analytics uses Google’s measurement script after permission; no advertising tracking or session replay is installed. The analytics choice and an optional internal-testing exclusion are stored in this browser. Local event receipts prevent duplicate analytics observations without sending account identifiers to Google. Profile answers are not saved in durable browser storage by the app. Authentication providers may use necessary session storage and security logs.
This pilot is limited to people 18 and older. We do not intentionally collect information from children. We may update this policy as the service changes, and will communicate material changes through the application or necessary account messages.
Operator: [Draft: operator legal identity required]. [Draft: privacy contact email required before public launch]. This is draft beta copy and requires founder and counsel review.