Privacy

The free analyzer stays in your browser

ViewPulse parses the history file you select and computes the free analysis in your browser. The free path does not upload your filename, raw export, watched URLs, titles, channels, keywords, or history-derived values. Games are optional. Account AI is a separate, explicit signed-in action. An account is not required for the free analyzer.

Information we access, collect, store, and use

The information involved depends on the feature you choose. ViewPulse does not automatically read your YouTube account or viewing history.

• Browser-local Takeout data. When you choose a Google Takeout watch-history file for free analysis, the browser reads its filename, watched URLs, video titles, channel names, source type, and timestamps. These values are used to create the free analysis on your device and are not uploaded by the free path.
• Account and session data. If you create an account, Clerk processes identifiers such as your user ID, email address, profile details you provide, session state, device and security signals. ViewPulse uses them to authenticate you, isolate your workspace, recover entitlements, and prevent abuse.
• Payment data. Stripe processes billing contact, payment-method, customer, product, transaction, subscription, tax, refund, and dispute information. ViewPulse receives bounded customer, transaction, status, and entitlement identifiers but not full card details. Checkout does not receive watch history.
• History you explicitly submit. Hosted Memory and one-time workflows store only the normalized titles, channels, timestamps, source type, and recoverable public YouTube video IDs that you separately consent to submit. ViewPulse uses them for the requested analysis, Chat, One-click tools, MCP retrieval, support, cost control, and abuse prevention. The raw Takeout archive is not stored.
• Public YouTube API Data. For selected public videos and channels, ViewPulse may access and temporarily store the identifiers, titles, descriptions, thumbnails, publication dates, durations, public engagement statistics, public comments, channel metadata, search results, and public transcript evidence described below. This data is used only to verify sources and produce the user-requested result.
• Prompts, outputs, and evidence. Account AI and paid workflows process the question or instruction you submit, bounded history evidence or aggregate analytics needed for that task, model output, measured usage, and the resulting artifact. They are used to answer the request, enforce evidence and safety contracts, settle credits, and deliver or recover the result.
• Operational data. ViewPulse and its hosting and security services process ordinary request metadata such as IP address, user agent, request time, route, request or operation ID, error state, and security events where needed to operate, diagnose, rate-limit, and protect the service.

Premium workflows are separate and explicit

AI Chat + MCP, Brainrot Control, Knowledge Map, Custom Recommendations, Psychological Review, and Memory Boost are distinct hosted workflows. Checkout sends no watch history. After a verified purchase, ViewPulse asks for separate, versioned consent before you manually submit normalized history. ViewPulse does not connect to or automatically synchronize your YouTube account, and does not store a raw Google Takeout archive for these workflows.

Psychological Review provides framework analysis only. Memory Boost is a separate workflow with 50 questions and 50 reflections based on retained transcript evidence. Neither workflow establishes measured personality traits or learning outcomes.

A paid submission contains normalized titles, channels, timestamps, source type, and recoverable public video IDs when present. It is encrypted with an owner- and order-bound key and isolated by account. Every valid normalized event may participate in deterministic aggregation. Any content-level AI review is bounded representative evidence: ViewPulse does not claim that a history entry proves watch completion, attention, intent, or effect, or that every video was transcribed.

Premium checkout remains disabled unless authentication, exact Stripe pricing, signed webhooks, encrypted storage, worker execution, provider cost controls, artifact delivery, retention, and deletion gates pass for that product. Account AI uses the separately described, credit-bounded OpenRouter route; each paid workflow remains independently gated until its model and discovery checks pass. ViewPulse will not silently send history to an unlisted fallback provider.

Retention, recovery, and deletion

Each paid submission carries a deletion deadline; the current workflow contract limits submitted normalized history to 30 days. Activation also requires an executable process that deletes expired source material and handles artifact recovery according to the terms shown before submission. Billing, entitlement, revocation, and bounded audit records may be retained as needed for purchase recovery, disputes, fraud prevention, and legal record-keeping, but they do not contain watch-history content.

Whole-account deletion and key retirement must prevent future access to encrypted customer content. Backups may retain already encrypted, cryptographically erased ciphertext only for a separately bounded backup window; ViewPulse does not claim physical per-user removal from a monolithic backup. Send a deletion request to the address below and never include watch-history content in the request.

YouTube API Services and public API data

ViewPulse uses YouTube API Services for bounded premium-workflow evidence and recommendation discovery. It requests public, non-authorized API data with a deployment API key: video and channel identifiers, titles, descriptions, thumbnails, publication dates, durations, public engagement statistics, public comments, channel metadata, and search results. ViewPulse uses that data only to verify sources, build the result the user requested, and explain evidence shortfalls. It does not use YouTube OAuth, ask for a YouTube password, act on a YouTube account, or retrieve a user's private YouTube data.

Public YouTube API data included in a submitted workflow follows the same encrypted, account-isolated controls described above. It is deleted or refreshed within 30 days. If a user asks ViewPulse to delete stored API data, ViewPulse deletes it as soon as possible and no later than seven calendar days; this removes ViewPulse's copy and does not delete anything from YouTube. Deletion requests go to Yev@self-degree.com and must not contain watch history or artifact content.

Google handles information under the Google Privacy Policy. Google account permissions can be reviewed on the Google security permissions page. ViewPulse currently requests no Google account authorization, so no ViewPulse OAuth grant should appear there.

How information is processed and shared

Access inside Self-Degree is limited to operating, securing, supporting, and improving the feature the user selected. ViewPulse does not sell watch history and does not share it for advertising or ad personalization. The following external processors receive only the information needed for their role:

Clerk provides account authentication for premium checkout, workspace access, entitlement recovery, and named MCP connections. Stripe handles one-time and subscription payments. Stripe receives product, price, customer, and bounded transaction identifiers, but not watch-history or artifact content.

OpenRouter and approved model hosts process task-bound prompts, bounded evidence, aggregate analytics, and model outputs. Eligible hosts are selected through OpenRouter's provider routing for Chat/MCP synthesis, transcript analysis, staged evidence analysis, premium generation, and evaluation. GLM is used for the approved text route. Different stages may use different approved models. Text tasks may include bounded private watch-history context and derived evidence needed for the workflow you explicitly request. OpenRouter routes may also use OpenAI GPT-6 Luna hosted by Microsoft Azure for Memory Boost writing, field repairs, source-grounding, and duplicate-activity checks, and for Knowledge Map concept summaries, graph drafting, bounded editing, and source-grounding checks, and for Custom Recommendations source-isolated scoring and recommendation rationales. Recommendations source analysis and search-query planning remain on GLM. These Azure routes disable response caching and automatic prompt caching and use the same no-data-collection and zero-data-retention routing requirements. These checks do not create an independent guarantee of factual accuracy. Other OpenRouter routes may include Google-hosted text models only when separately approved for that task. This text-processing disclosure does not make an experimental model or workflow available for purchase. Automated generation, verification, and separate-session evaluation may share a model family. These checks are fallible, are not necessarily independent-model validation, and do not establish a validated personality test. Text-generation, evaluation, and public-video requests routed through OpenRouter require OpenRouter's no-data-collection and zero-data-retention routing controls. Separately configured native public-video requests exclude private watch-history context. ViewPulse does not silently route private tasks to an unapproved model host. Text processing through OpenRouter is separate from the public-only opening-clip feature below.

Optional image generation. When this feature is enabled, clicking Generate image sends only the image prompt you review and edit to OpenRouter and OpenAI (GPT Image 2.5 Flare). Your watch-history archive, aggregate analytics, and account identifiers are not included in that provider request. Do not include personal information in image prompts. Unlike the text routes above, this image route uses the providers' standard retention policies and is not represented as zero-data-retention. ViewPulse keeps the returned image only in your current page session; download it if you want to keep it. Bounded usage and credit records are retained for billing, without storing the image or prompt in those records. This optional route does not change the privacy restrictions for Chat, Hosted Memory, MCP, or other workflows.

Supadata is the temporary public-video evidence and discovery service while ViewPulse completes its YouTube API review. It receives public video IDs or URLs and bounded public search inputs needed to return public metadata, transcript evidence, and discovery candidates. It does not receive a raw Takeout archive, YouTube login credentials, private YouTube account data, or Stripe payment data. No transcript check or AI transcription runs before the initial payment for a new workflow purchase. After verified payment and your separate history submission consent, ViewPulse sends public video identifiers to Supadata to retrieve existing captions only. Videos with missing captions are skipped and their count is disclosed after the native-caption pass. Private or unavailable videos, including ongoing livestreams, are disclosed separately and may not be processable. Successful empty transcripts (no speech returned) are disclosed and excluded, not offered for AI transcription. Provider failures are not treated as proof that captions are missing. We retain encrypted source availability results and transcript checkpoints within the workflow's deletion deadline, for up to 30 days.

Eligible videos with missing captions can be included in an optional add-on, when available. The displayed price covers transcription and report regeneration; it requires separate approval and payment. A separate updated report is produced if processing succeeds; the original report is not replaced. For these approved videos, Supadata auto mode uses existing captions when available or AI-generated speech transcription when needed. The response may not identify which method produced a transcript. Transcription can contain errors and cannot make unavailable videos accessible. Legacy purchases retain their previously approved transcription scope.

Google Gemini, when the opening-clip feature is enabled for Brainrot Control, receives a public video URL, its duration, and instructions to inspect only the first minute (or the entire video if shorter) at one sampled frame per second. It does not receive your Takeout archive, watch dates, account identity, or other private history context through this feature. Requests disable stored interactions; this does not mean that all provider operational or abuse-monitoring retention is disabled. ViewPulse retains its encrypted workflow checkpoints for up to 30 days. Other report workflows do not use this opening-clip inspection.

YouTube and Google receive public video or channel identifiers and public search requests when the YouTube API route is used. The privacy-enhanced YouTube player is loaded only when a user opens a cited-video preview. ViewPulse does not send Google a Takeout archive or request YouTube OAuth access.

PostHog is optional and loads only after explicit analytics consent. It uses a random anonymous ID and bounded funnel events. Profiles, autocapture, session replay, query strings, click IDs, and history-derived text are disabled or rejected. Withdrawing consent stops the current analytics generation.

Google Ads is optional and loads only after separate conversion-measurement consent. For a verified purchase it receives the conversion destination, purchase value, USD currency, and a one-way 64-character transaction digest used for deduplication. It does not receive watch-history content, video or channel names, report text, or the raw Stripe session identifier. Ad personalization and Google analytics storage remain denied. Withdrawing this consent updates Google consent mode to denied.

Account AI and provider credentials

ViewPulse does not request or store your OpenAI, Gemini, or OpenRouter API key. Account AI calls use a deployment-managed OpenRouter credential on the server and are bounded by your authenticated credit allowance. Provider usage is measured for cost control.

Free chat sends only bounded title, channel, and date rows selected by the browser-local index for that question. With an active AI Chat + MCP plan, you may instead explicitly import a normalized archive to Hosted Memory. Its history text is encrypted per account at rest; no semantic vectors are stored. Server-side retrieval supplies at most eight cited rows to each chat request. One-click tools send aggregate analytics rather than the complete raw archive. The public analyzer remains local unless you deliberately continue to an account workflow.

Hosted Memory is retained through the paid-through date plus a 30-day recovery window and is deleted after that boundary. You can delete it earlier from the authenticated account flow. MCP tokens are stored only as digests, are read-only, share a bounded monthly quota, and can be revoked individually.

Cookies, browser storage, and similar device technologies

ViewPulse accesses and stores information on your browser or device, and allows the providers named below to do so. Essential Clerk cookies and similar session storage keep you signed in and protect authenticated requests. Stripe may place or recognize cookies and similar technologies on its hosted checkout and billing pages for payment, fraud-prevention, and security purposes. These essential technologies are not controlled by the optional measurement choice.

ViewPulse local storage or session storage may retain your privacy choices, a random anonymous analytics identifier, first-touch attribution for up to 90 days, game progress, and a short-lived opaque handle used to move a browser-local analysis between pages. Provider API keys, raw Takeout files, titles, and channels are not placed in those browser records. Clear site data to remove them sooner, although doing so can sign you out or reset saved choices.

Optional PostHog product analytics and Google Ads conversion measurement load only after the corresponding consent. PostHog is configured with memory-only persistence and no autocapture or session replay. Google Ads may place, access, or recognize cookies or similar technologies for the consented purchase-conversion measurement; Google Analytics storage and ad personalization remain denied. A privacy-enhanced player from youtube-nocookie.com is loaded only after you choose to preview a cited video and may process device or network information under Google's privacy terms. You can change optional choices at any time with Privacy choicesin the site footer.

Contact

ViewPulse is operated by SELF DEGREE EDUCATION EDUCATIONAL TECHNOLOGIES - FZCO, licensed by the Dubai Integrated Economic Zones Authority under Trade Licence No. 76863. Its licensed business address is Premises DSO-IFZA, IFZA Properties, Dubai Silicon Oasis, Dubai, United Arab Emirates. Send privacy and deletion requests to Yev@self-degree.com. Deleting entitlement records may end purchase recovery. Never send watch history, exports, or artifact content with a request.