Privacy Policy
How Taskora handles your account, conversations, files and connected services — and how you can manage your information.
Last updated: 13 October 2026 · Draft for review
Effective date: Pending operator approval — preview only
Review draft, not yet effective. Operator identity, grievance handling, retention, provider safeguards and consent decisions remain unresolved. Account-deletion processing is currently disabled and the data export is partial.
1. Introduction and scope
Taskora is an AI-assisted service for conversations, images, documents, web research, saved information, lists and reminders, with optional Google connections and paid subscriptions. This policy covers visitors to the website and users of the signed-in application. Features depend on your plan, available credits, permissions and the services currently enabled; not every feature is available to every user.
Operator confirmation required: Taskora’s legal operator name, registered or contact address, operating country and any required grievance officer have not been confirmed. The name Taskora identifies the service, not a verified incorporated company. This preview is not yet an effective legal notice. The service serves users in India; applicable-law and international-user disclosures require operator and legal review.
2. Information we collect
The categories below depend on what you do in Taskora. Content may contain information about other people; provide only information you are entitled to share and that is needed for your request.
- Provided by you: email and password for email-based registration, display name, optional profile photo, prompts, attachments, feedback, saved memories, lists, reminders, notification choices and support messages. A chosen or detected time zone helps schedule reminders.
- Generated through use: account and conversation identifiers, messages and AI responses, conversation context summaries, document excerpts, generated images or PDFs, action proposals and completion states, search citations, timestamps and read/unread notification states.
- Account and technical records: verification and sign-in states, session tokens, browser or user-agent information, session activity and revocation timestamps, IP-related security and rate-limit records, request references, error codes, security events and audit records.
- Received from providers: permitted Google identity information at sign-in; when connected separately, Google account identifiers, email address, granted permissions, access and refresh tokens, and data returned by requested Google actions. Payment providers return subscription, payment and refund references and status.
- Plan and operational records: plan access, billing dates, amounts and currency, payment/refund states, offer eligibility or redemption where used, credit balances and reservations, model and tool usage, token counts, cost records and duplicate-request safeguards.
- Browser and hosting information: functional browser storage, observed hosting/security cookies, font requests, and hosted performance, page-use and error diagnostics described below. The purpose of the observed __dpl cookie and a complete hosted collection and retention inventory still need operator review.
3. How we use information
We process information to register and authenticate accounts, manage profiles and sessions, respond to your requests, preserve saved conversations, process supported files and images, and perform requested research. Saved memories can provide relevant context when the feature and your preference allow it.
We also use it to operate lists, reminders and notifications; perform authorized Google actions; verify payments and manage subscriptions, credits and limits; prevent abuse; investigate errors; and handle support or privacy requests. Security and billing records support accountability and obligations that apply to the operator. The precise legal grounds and mandatory recordkeeping duties must be confirmed before finalization.
Service notices can be shown in the application. This audit did not establish an implemented marketing-mail programme or marketing subscription control. It does not promise promotional emails or automatic email notice of every service change.
4. AI and service providers
The currently enabled AI provider is OpenAI. Text requests can include your prompt, relevant conversation context, selected saved information, file excerpts and images when needed. Image generation uses Taskora’s configured direct OpenAI image service. OpenAI also supplies the hosted web-search feature. This audit found no active alternative AI provider in the configuration reviewed.
Speak to type is optional and uses the speech recognition built into your browser. Your microphone is used only after you tap Speak to type and allow it, and recognised words are only placed in the message box; nothing is sent until you press Send. Your browser or device maker may process your speech on its own servers rather than on your device: for example, Google Chrome and Microsoft Edge use their providers' online speech services, and Apple Safari may process speech on the device or on Apple's servers. This depends on your browser, device and their settings, and is covered by that provider's privacy terms; Taskora does not control how the provider processes your speech. Taskora does not receive, record or store your audio, only the text you choose to send.
The text and hosted-search requests inspected ask the provider not to save an API response as persistent response state. This is not a promise of zero retention: abuse monitoring, exceptions and other processing depend on the exact provider service and account settings. Taskora’s account-specific provider retention and contractual safeguards have not been verified. Consult provider privacy and API terms before submitting sensitive material.
Lovable Cloud provides application data, authentication and private file infrastructure. The hosted website also uses Cloudflare-related delivery/security services and hosting diagnostics. Google provides authorized integrations and external fonts; Firebase Cloud Messaging supports optional push delivery; Razorpay processes payments. File extraction and PDF processing are implemented within Taskora’s service infrastructure rather than through an identified separate document-extraction API.
No Taskora-operated model-training pipeline using user conversations was found in the implementation. This does not establish the operator’s complete business practices or guarantee that every provider excludes model improvement under every configuration.
5. Conversations and saved content
Saved chats retain your messages, replies, relevant action metadata and identifiers so you can return to them. Taskora may derive a bounded context summary or retrieve relevant older messages from the same chat to answer follow-ups. Saved memories and lists are separate records; their settings and removal controls are available in the application.
User-facing reads are restricted to the owning account and an active session. Server-side authorization checks protect supported changes and downloads. These are access controls, not end-to-end encryption or a guarantee that no authorized operator or infrastructure administrator can ever access data.
Deleting a chat hides it and clears message and reply text, removes associated saved memories/lists and requests removal of its attachments. Some identifiers, action metadata, execution, usage and audit records remain. Generated-file cleanup is separate and can be delayed; deleting a chat is not a promise that every associated record or backup disappears instantly.
Secure Chat (previously called Temporary chat) is kept out of your chat list, search, Archived, Library, memory, summaries, ratings and offline copies on your device. Its messages, replies and attachments are stored on our servers for up to 24 hours from the chat's creation so replies can be produced. From that point the chat is refused on every request, even before cleanup runs; hourly maintenance then clears the message and reply text and marks attachments for removal, and file bytes are removed on later runs.
After Secure Chat text is cleared we keep the identifiers and timestamps needed to run the service, and the usage, cost and credit records needed for billing and abuse prevention. Clearing message content is not the same as deleting these records, which follow the financial-retention rules described below. Database backups may contain copies of Secure Chat content for a period we have not yet confirmed.
Each Secure Chat message is sent to an AI provider to generate the reply, under the same processing as ordinary chats. We have not verified whether, or for how long, the provider retains these requests. Secure Chat is not end-to-end encrypted and is not anonymous: it is linked to your signed-in account.
6. Images and uploaded files
Image-generation prompts are sent to OpenAI to produce a new image. The result is saved in private application storage with information such as its type, size, dimensions and associated chat. Viewing or downloading it uses an owner-checked temporary link, currently valid for two minutes. Link expiry does not delete the stored image or a copy you downloaded.
Image inputs support JPEG, PNG and WebP. Current global upload controls allow up to 10 MiB per image and up to 10 images per request, subject to model availability and other checks. Images may be resized and re-encoded before storage and AI processing; conversion can fall back to validated input bytes if optimization fails. Uploaded-image viewing links currently last five minutes.
Supported document inputs are PDF, DOCX, XLSX, CSV and TXT. The current global file baseline is 10 MiB per file and five files per message, with additional page, extracted-text and spreadsheet limits; your plan can change effective limits. File type and contents are checked. Image-only scanned PDFs are not treated as readable text by the current extraction implementation.
Uploaded document bytes and file metadata are stored privately. Extracted text is stored separately in access-restricted records and relevant excerpts can be sent to the AI provider. Generated PDFs are also stored privately and use short-lived download links. Supported PDF editing processes the supplied document without making every editing action an AI request.
Unsent or orphaned uploads are currently eligible for cleanup after 24 hours. Chat deletion and temporary-chat expiry also trigger attachment cleanup, and deleting extracted attachments clears their text records. Cleanup timing depends on processing and outstanding work. No fixed maximum retention for all saved files or generated images has been established; see Retention and deletion.
7. Web search and third-party pages
When web search is used, relevant request content and the search query can reach OpenAI’s hosted search service. Returned source links, titles and citations can appear in your stored reply; tool counts and costs are recorded for usage. There is no separately configured search-API provider in the implementation reviewed.
Website research can also retrieve public pages and source metadata from external sites. Those sites receive network requests from the retrieval infrastructure; domain lookups use an external DNS service. Relevant retrieved page text may be included in AI processing. Opening a source link yourself sends a browser request to that website.
Searches are not guaranteed anonymous, and external sites and search services have their own privacy practices. Taskora does not control those practices. Search availability depends on the plan, model, credits and enabled tools.
8. Lists, reminders and notifications
Taskora’s own lists and reminders can store item text, completion state, due times, time zones, recurrence, scheduling and delivery status. Related notifications store message text, timestamps and read status. Some requests are handled directly without AI; content included in a chat or an AI-assisted request may still be processed as conversation content.
Optional browser/push alerts require your browser permission and a supported setup. Push delivery uses Firebase Cloud Messaging and stores device-registration tokens, device/browser metadata and preferences. Notifications may contain a generic alert or reminder details depending on your settings. Choose carefully on shared devices or visible lock screens; notification delivery is not guaranteed.
You can manage supported reminders and application alert choices, and revoke browser notification permission in your browser or device settings. Google Tasks is not an available integration: Taskora’s lists and reminders are distinct from Google Tasks.
9. Optional Google integrations
Google sign-in does not itself authorize Gmail, Calendar or Drive access. Connecting these services from Settings requires a separate authorization flow. Access depends on the services selected and the permissions Google actually grants; configured credentials do not establish that every Google API or account can currently connect.
Gmail: the requested permissions allow reading/searching mail and sending mail. Supported reads can return sender/recipient information, subjects, snippets and message content. Sending or scheduling requires your explicit Send or Schedule confirmation on a draft. The integration does not implement deleting or modifying existing mail. Scheduled-send records retain recipient, subject, timing and a protected copy of the body.
Google Calendar: the requested event permission supports listing/reading events and creating, moving or deleting supported events. Event details can include title, description and times. Changes are proposed and require confirmation before execution; an AI-generated proposal alone does not make the change.
Google Drive: the requested permission is read-only. Supported features search/list file metadata and read/export supported content or excerpts, including supported Google documents and spreadsheets. No Drive write, upload or deletion action is implemented by this connection.
Taskora stores the connected Google account identifier, email address, granted permissions, connection state and token-expiry information. Access and refresh tokens are encrypted server-side and are not returned to the browser as ordinary account data. Connection/action audit metadata is separate from these tokens.
Retrieved Google content is sanitized for display and the reviewed chat-context path replaces integration replies with a placeholder rather than passing the raw retrieved content back to AI. Your own request, an email draft you provide, and relevant draft context may be sent to OpenAI for drafting or rewriting. Do not interpret this as a claim that no information associated with Google actions ever reaches AI.
Use Disconnect in Settings to remove the saved connection and its tokens. Taskora attempts to revoke authorization at Google, but revocation can fail; you can also remove access in your Google account’s third-party connection controls. Disconnecting does not erase prior chat excerpts, draft/scheduled-send records or audit metadata, or undo emails/events already created. Pending actions may fail without a connection. Ask support about stored copies you want removed.
10. Payments, subscriptions and refunds
Taskora’s implemented payment processor is Razorpay. The integration is enabled and has a recorded verification, but this privacy audit did not perform a checkout or independently verify the current live credentials and complete production readiness. Payment availability must be determined by the actual checkout; this policy does not promise that live payments are currently available.
Taskora records customer, subscription, plan, payment and invoice references; test/live mode; amount and currency; billing dates and states; recorded fees/tax; and refund status or amounts. Offers can also produce eligibility, redemption and bonus-credit records. Payment verification and signed webhook events reconcile these records; browser checkout success alone does not activate a plan.
The implemented checkout uses Razorpay’s payment interface, not a Taskora card-entry form. Taskora’s billing records do not store full card numbers or card security codes. Razorpay can process payment, identity, device and fraud-prevention information under its own privacy practices. Do not send card details, passwords or one-time codes to support.
Verified test-only webhook deliveries are kept separately as an event name and server timestamp, without their payload, and do not update live billing. Live webhook processing records delivery references and processing states for verification and reliability.
Account deletion is not cancellation or a refund. Review Billing and the Terms before changing a subscription. The owner selected a refund window but has not confirmed its duration, eligibility or exclusions; no refund deadline, approval promise or processing time is invented in this policy.
11. Cookies, storage and diagnostics
Taskora stores sign-in state in browser storage. Local/session storage can also hold appearance and colour choices, typing and alert preferences, sign-in destinations, recovery markers and pending preference updates. Appearance follows your device until you explicitly choose otherwise. A homepage draft can be carried through sign-in for up to 30 minutes; expired drafts are rejected when read, not necessarily removed from an unused device at that instant.
The optional offline experience can save a recent, account-specific chat snapshot locally. The application clears that snapshot on sign-out. Its service worker caches application resources for availability. Browser caches and copies on your device are separate from server-side deletion; take care on shared devices.
A read-only visit to the live site observed a Cloudflare bot-management cookie (__cf_bm) and a cookie named __dpl whose purpose and retention have not been established. The page also loaded hosting performance and event-diagnostic scripts. The inspected event script supports page views, navigation/referrer information, engagement, technical/browser metadata, errors and session identifiers; it also contains configurable session-replay functionality. Loading this script is not proof that replay is enabled for every visitor or that its masking excludes all private content.
Google Fonts is requested when pages load, exposing ordinary request information such as the network address and browser headers to Google. Razorpay’s checkout script is loaded when payment is initiated, and Firebase scripts/services can be used for enabled push alerts. No app-authored advertising pixel was found, but this is not a guarantee that hosting or provider technologies never collect analytics.
You can clear or block cookies/storage in your browser, sign out, disable offline/alert choices where available and revoke notification permission. These actions may remove local drafts, disable features or require another sign-in. There is no implemented optional-analytics consent selector in the reviewed application. Operator review is required to establish hosted tracking/replay configuration and any jurisdiction-specific prior-consent requirements; this preview does not claim those requirements have been fulfilled.
13. International processing
Providers and their infrastructure may process information outside India or your country of residence. A regional connection indicator in the infrastructure metadata is not enough to establish where all databases, storage, backups, logs, AI or notification processing occur.
Operator confirmation required: establish the actual hosting and processor regions, subprocessor list, relevant contracts and safeguards, and the lawful conditions for applicable transfers. This audit does not verify an India-only location, adequacy determination, standard contractual clauses, certification or a provider’s account-specific transfer arrangement.
14. Retention and deletion
Retention differs by category. The periods below are verified settings or code behavior at this revision, not a universal deletion promise. Categories without a documented maximum need a purpose-based retention schedule established and reviewed by the operator before finalization.
Settings includes an account-deletion request with typed confirmation and a currently configured seven-day grace period. Requesting deletion records a request and marks the profile; it does not delete data, cancel a subscription or close the account. A request can be cancelled while its status is pending. Automatic account-deletion processing is disabled, and no account-deletion job was found in the reviewed schedule. Passing the end of the grace period therefore does not erase anything: the request stays recorded as pending, and Settings states that nothing is erased automatically. A recorded request is not completed erasure, and Taskora does not currently send a completion confirmation because no deletion is completed automatically.
The disabled processor is not a verified complete-erasure solution. Its reviewed cleanup does not explicitly remove conversations or their messages, and those conversations do not cascade when the account is removed. Other usage, credit, audit and security records have separate lifecycles. Storage cleanup coverage, error recovery, Google authorization revocation and lawful billing-record preservation need engineering and legal review before any activation. Contact support to request review, not as a promise that a complete manual-erasure process is already available.
- Account/profile data: retained while the account exists; the request workflow and disabled processor are described above. No verified maximum applies to every residual account-related record.
- Saved chats and attachments: retained until removed or subject to feature cleanup. Chat removal clears text and hides the chat but retains some identifiers/metadata. Secure Chat content becomes unavailable 24 hours after creation and its text is cleared by the next hourly cleanup (timestamps and usage/credit records remain; backup retention unconfirmed); orphaned uploads are eligible after 24 hours. Physical file cleanup can follow later.
- Generated images and PDFs: private stored outputs remain associated with their chat; a cleanup routine targets deleted or unavailable chats. A complete recurring purge and maximum retention have not been verified. Download-link expiry is not file deletion.
- Google data: Disconnect removes the saved connection/tokens and attempts provider revocation. Previous chat content, action/draft/scheduled-send records and audit events have separate lifecycles; no single maximum is documented for them.
- Billing, credits and transactions: stored for subscriptions, accounting, reconciliation and support. The owner has decided that historical payment, refund, subscription, credit, usage, cost and audit records are to be preserved rather than erased or rewritten when an account is deleted, subject to applicable retention requirements. The safeguard that enforces this has been prepared and tested in isolation but is not yet applied, so it is not described as active. Statutory retention periods and any later minimization of identifiers require legal review; no fixed period is established.
- Security, audit, usage and diagnostics: expired rate-limit windows are eligible for cleanup after an additional configured 24 hours. A 365-day usage setting exists, but an enforcing purge was not found, so it is not a verified usage-deletion deadline. No maximum was established for append-only audit/security events, credit records or hosting diagnostics.
- Data exports: files are available for seven days. Every download, including a retried request, checks that the file is ready and unexpired before a link is issued, and each signed link lasts at most 15 minutes. A cleanup routine is designed to remove expired or failed export files from storage; its scheduled operation has not yet been verified.
- Backups: backup existence, locations, schedules, retention and post-deletion removal have not been verified. Third-party copies and copies downloaded by recipients or retained on your devices have their own handling.
15. Security measures
Verified application protections include authentication and verification checks, session revocation and inactivity controls, account-based access restrictions, server-side authorization and ownership checks, rate limits, private file storage, protected integration credentials, and security/audit records. Downloads are issued through owner-checked, time-limited links.
These measures reduce risk; they do not guarantee perfect security, end-to-end encryption or protection against every incident. Keep credentials and download links private. Report suspected unauthorized access to support and include only the reference needed to investigate.
16. Your rights and choices
Available controls include editing your display name/photo in Profile; managing appearance, time zone, supported alerts and Google connections in Settings; managing saved memories/lists and reminders; deleting chats; and requesting an account export or deletion. There is no verified self-service email-change control or comprehensive privacy-case tracking portal.
‘Download my data’ currently builds a limited JSON export of account email/identifier, selected profile fields, roles, session metadata, notification records (including their titles and message text) and deletion-request records. It does not include conversations, attachments, generated images or PDFs, memories/lists, reminders, Google records, full billing/credit history, usage/audit records or push-device records. The file states which categories it includes; a category that could not be read is omitted and the file is labelled partial rather than shown as empty. Each download link lasts at most 15 minutes. Contact support to request review of missing information; a comprehensive manual export process has not been verified.
Depending on applicable law, you may request access, correction, erasure, withdrawal of optional consent/permissions, restrictions on processing or complaint/grievance handling. Request-specific eligibility, exceptions and deadlines depend on the law in force and the operator’s circumstances; this preview does not claim every right applies everywhere or establish a new statutory deadline.
Email support@taskorachatbot.com for manual requests or privacy complaints. State the account email, what you are requesting and enough context to identify the issue. Ownership may need verification; never supply your password or one-time code. Removing optional permissions stops the relevant connected feature but does not automatically remove prior records or information a provider independently retains.
Indian legal review: commencement of the Digital Personal Data Protection Act and final rules is phased. As of 13 October 2026, the main consent, children’s-data and data-principal-rights provisions are scheduled for later commencement, not all already operative. Existing Indian privacy and consumer requirements may independently apply. Qualified review is needed now and again as provisions commence; no DPDP or international compliance certification is claimed.
17. Children and age restrictions
Operator decision required: a minimum-age policy, applicable child-data rules and any parental-verification process have not been established in the implementation or finalized Terms. This preview does not invent an age threshold or claim an age gate exists.
If you believe a child’s information has been provided without appropriate authorization, contact support with the information needed to locate it. The operator must establish a reviewed investigation, consent and removal process, taking account of current law and upcoming Indian requirements, before this section is finalized.
18. Third-party websites
Links, citations, connected accounts, payment interfaces and external websites are governed by their own privacy notices and terms. A Taskora link does not extend this policy to the destination. Review the destination’s practices before providing information or downloading material.
19. Policy changes and dates
This page shows the latest revision date. Its effective date remains pending because this is an unpublished review draft with unresolved operator and legal decisions. The operator must approve a final effective date and an appropriate material-change notice process before publication.
Material changes should be reflected on this page and communicated as required by applicable law. No automatic policy-change email workflow was verified; this draft does not promise that every user will receive an email. Review this page together with the Terms and current account/checkout information.
20. Contact and grievance handling
The verified support and privacy contact is support@taskorachatbot.com. Use it for privacy questions, access/correction/deletion requests and complaints. Avoid unnecessary sensitive attachments and include a request or payment reference where relevant.
Operator confirmation required: legal operator name, contact/registered address, operating country, designated grievance contact and any required escalation procedure. A named officer, phone number, address or statutory response commitment is not fabricated here. The operator must confirm whether a separate privacy address is needed and establish a legally reviewed grievance process.