Privacy Policy

Updated October 8, 2026

Day ’N Deed is the new name for Sovereign Isla. The public App Store release is coming soon. This policy describes the Day ’N Deed release in testing. Older beta builds may have fewer features.

Who operates the app

Day ’N Deed is operated by 1Fisher Group LLC. Contact info@1fishergroup.com for privacy questions.

Information we use

• Account identifiers, names, email addresses and information supplied by your sign-in provider.
• Content you save: tasks, lists and list items, team membership, categories, task comments, team conversations, notes, projects, goals, meeting transcripts, summaries, action items, selected contact details, vocabulary and schedule preferences. In the enhanced release in testing, this also includes private commitments and their reviewed source/history, task estimates and prerequisites, selected calendar block references and times, and goal success measures and check-ins.
• Audio submitted for transcription and text or relevant task and meeting context submitted when you request AI assistance.
• Subscription identifiers, purchase status, expiration dates and usage counters. Apple processes App Store payments and Google processes Google Play payments; we do not receive your full payment card details.
• Technical connection and service information used to operate accounts, secure the service and diagnose failures.

Your choice before AI sharing

Optional AI processing is off until you explicitly allow it in the app. Before the first request, the app explains which personal data is sent, who receives it and why, and offers a choice to allow or decline. Reading this policy, signing in, buying a plan, or granting microphone or calendar access does not grant permission to send content to an AI provider. Manual tasks, lists, calendar views and private journal writing remain available without optional AI.

Open Profile & settings → AI & Privacy → Allow AI processing to turn optional AI off. This stops future AI assistance requests and clears the saved permission for speaker transcription. You can review this policy from that section. Withdrawing permission does not recall data already sent or delete saved results; deletion choices and provider retention are explained below. Message-safety processing has a separate permission because it is required to send shared messages.

What is sent to OpenAI

OpenAI receives content through Day ’N Deed’s authenticated backend for these purposes, after the applicable in-app permission:

• AI assistance: your prompt or dictated text and selected tasks, list names and items, commitments, project details, goals, meeting transcripts, summaries or relevant passages, together with relevant approved memory and saved spelling terms. These support task and list extraction, planning, summaries, questions and follow-up drafts. Content can include names, contact details or other personal information you put into those records.
• Daily and weekly briefs: selected saved work and meeting-preparation context, tasks and commitments, and titles and times from selected calendars. Weekly briefs use the selected day, week or month. Calendar notes, locations and private journal entries are excluded from those briefing requests. Building or refreshing a factual brief on the device does not itself send its contents to OpenAI.
• Basic cloud transcription: the recording you choose to transcribe, with relevant saved spelling terms, to turn speech into text. Recordings can contain the voices and personal information of you and other people. OpenAI may also receive the recording as the disclosed fallback for speaker transcription.
• Natural narration: the selected meeting passage or displayed brief, plus the chosen narrator voice, to create audio. This has a separate in-app choice. Replaying prepared audio sends no new text. Included voice previews do not send your meeting content.
• Journal reflections: only the entry or recent entries identified in the journal’s separate confirmation, when you choose that AI action. Journal text is excluded from general Ask AI searches, daily and weekly briefs and meeting questions.
• Shared-message safety: the task comment or team-message text you choose to send, to detect potentially harmful content before delivery. This requires separate acceptance of the message-safety disclosure and community rules; it can operate while optional AI assistance is off and uses no AI credits. Declining keeps the message unsent.

What is sent to ElevenLabs

Identify speakers sends your selected meeting recording and relevant saved spelling hints to ElevenLabs to transcribe speech and label different voices. The recording can contain other people’s voices and anything said in the conversation. A separate prompt names ElevenLabs and the OpenAI fallback before this upload is permitted. Speaker labels distinguish voices; they do not verify anyone’s identity. If ElevenLabs cannot complete the request, the recording may be sent to OpenAI for the same purpose. Saved transcripts may then be sent to OpenAI for a summary when you have enabled AI assistance.

Only record or upload conversations you are permitted to process and share. Do not use ElevenLabs speaker transcription for recordings containing anyone under 18: ElevenLabs prohibits submitting children’s voice data. You can decline cloud processing and keep the original recording for playback or a later transcription choice.

On-device alternatives

Supported native speech recognition processes speech on the device. For meetings, you can choose Apple’s on-device transcription on supported iOS 26 devices, speaker identification in the cloud, or to be asked each time. On Android, optional offline English transcription downloads an approximately 148 MB whisper.cpp speech model from Hugging Face after consent; recordings stay on the device during that transcription path. On-device transcription and system Device voice read-aloud do not send that audio or text to OpenAI or ElevenLabs and do not use a cloud allowance. Cloud fallback requires the applicable permission; declining preserves the recording. Review results because transcription and AI can misunderstand names, dates, commitments and speakers.

AI data protection and provider retention

We require third-party service providers that receive personal data on our behalf to provide the same or equal protection described in this policy, including limits on processing purposes, confidentiality, appropriate security safeguards and assistance with privacy requests. OpenAI and ElevenLabs process customer personal data under their applicable OpenAI data-processing terms and ElevenLabs data-processing terms. Those terms also require protections for subprocessors and assistance with security incidents. The provider-specific retention details below remain applicable.

Our app sends AI requests over encrypted HTTPS connections through our authenticated backend. Provider credentials stay on the server. Account access controls restrict access to saved records; private journal contents are excluded from general AI context. We send the content needed for the selected feature rather than automatically uploading your entire address book, calendar or journal. Information you include in a prompt, recording or saved record may still identify you or another person.

OpenAI’s API data policy states that API content is not used to train its models by default. Our text-generation requests disable stored responses. This is not a promise of zero retention: OpenAI may retain abuse-monitoring logs for up to 30 days, with longer retention for legal or safety reasons, and some processing may involve temporary application state. Its policy lists no content retention for the transcription and moderation endpoints we use.

ElevenLabs’ service terms permit service improvement and provide an account-level training opt-out; its business data-processing terms govern personal data processed on our behalf. Retention depends on the service and account terms, with exceptions for legal requirements and backups. We do not promise zero retention or that an enterprise-only protection applies to all requests. See its privacy policy for information it processes for its own purposes. Choose on-device transcription if you do not want recording data sent to these providers. Contact info@1fishergroup.com for questions or requests concerning content previously processed.

Your private journal

Journal entries are encrypted on the device and unlock through the device’s biometric or device-passcode prompt. The app does not receive your face data or device passcode. Journal text is not available to task collaborators or included in Ask AI searches.

Profile → Cloud & journal privacy offers optional end-to-end encrypted journal sync through Convex. Sync is off until enabled on that device. Entry contents, including titles, reflections, decision reasoning, review observations and linked-work references, are encrypted before upload. Decision entries preserve the recorded reasoning and keep subsequent reviews separately. These entries are excluded from daily and weekly briefs and meeting questions. Cloud records also contain account and entry identifiers, version numbers and deletion markers. The service does not receive the encryption key or recovery code.

Save your journal recovery code privately. You need it to link another device or recover after losing all linked devices; we cannot replace it. Pausing sync affects this device and keeps existing cloud entries and other linked devices. On an unlocked journal, sync runs when you open it, finish editing or request Sync; an internet connection is required. Conflicting edits may be preserved as recovered copies.

Optional cloud reflections require a separate choice to send the selected journal text to OpenAI. Creating a task requires reviewing the copied title. That task may later be shared, but its source journal entry remains private.

Shared work, lists and conversations

Members connect using invitations and choose what to share. Testing builds use private links or email-addressed invitations instead of visible codes. Task participants can see the task’s shared fields, assignments, completion status and comments. Team members can see that team’s work and conversation. List members can see shared items, quantities, sections and completion attribution. The app displays discussion audiences. List membership is managed separately from task assignment; private lists are not automatically shared by linking a task.

Only share content that those participants may see. Removing access prevents future authorized access but cannot remove copies another participant already saved. Your journal and its source text are excluded from collaboration.

Community safety

In build 19 and later, sending a task comment or team message requires accepting the community rules and message-safety processing. Message text is sent to OpenAI’s moderation service before delivery, separately from optional AI assistance. This does not consume your plan’s AI allowance. Private journal entries are excluded from message moderation.

Reports contain the reporter’s account identifier, reported member identifier, selected reason, explanation and a snapshot of the reported message or task, team or list context. Access is restricted to authorized support reviewers. The reported member cannot see the reporter’s identity. We retain report evidence and decisions to investigate abuse, including when the original message is deleted. Report records you submit are included in account deletion; reports submitted by others may remain for safety investigations. Blocking records control future sharing access. Contact us with privacy or appeal requests.

Device permissions

Microphone and speech permissions support voice capture. Calendar access displays events from calendar accounts enabled on your device, including Apple and Google accounts. In testing build 24 and later, Make time adds or updates a task time block in the calendar you choose. Blocks include the task title, scheduled time and a link marker, not private task notes or journal entries. People who can access a shared calendar can see its blocks. In the enhanced release in testing, task estimates, prerequisites, calendar-link markers and selected work-block times are stored with your account in Convex to reconnect and coordinate your planning across devices. Device-specific event identifiers and calendar-display preferences stay local. Preparing for a selected calendar meeting saves that event’s title, time and linking reference to your private meeting library. The app does not copy your whole calendar to its backend. Your operating system manages calendar-account synchronization. Creating or moving work blocks requires reviewing and applying the plan; pending updates can be reconnected without creating another event. Contacts access lets you select people for call, text and email tasks. Selected names, phone numbers and email addresses may be saved with tasks or your profile, and selected contact shortcuts (aliases) sync to your account through Convex to resolve future tasks. This does not upload the entire address book. Notifications provide reminders. You can revoke these permissions in your device Settings.

Providers and disclosures

Clerk provides account services; Convex provides the backend and cloud storage; OpenAI processes requested AI work and cloud transcription; ElevenLabs processes meeting audio for speaker identification after separate permission; Apple and Google provide their respective store purchase and device services; Expo/EAS supports app builds and distribution; Vercel hosts this website. Providers process information needed to deliver their services and may process data outside your country. We may disclose information when legally required or to protect the service.

We do not sell task or meeting content. The app does not contain an advertising-network SDK. This site does not currently run advertising pixels, nonessential analytics scripts or a mailing-list database. Contact links open your email app; messages you choose to send go to our support inbox.

Storage and retention

Saved cloud content remains associated with your account until removed or account deletion is requested. The app also retains local caches and preferences, including saved briefs and focus-session history. Recorded focus time and corrections are stored on the current device; goal success measures and progress check-ins sync with your account. Calendar reservations do not count as recorded effort. Lists cache content and pending edits separately for each signed-in member so previously loaded lists can be edited offline. Pending edits are checked against current access and item versions when reconnected. Signing out does not necessarily erase all local account files; removing the app removes its local data. Saved meeting audio stays on the recording device, separated by member account. Day ’N Deed does not automatically sync original audio between devices; another device can access the meeting text without the original recording. Depending on your iPhone backup settings, local audio may be included in a device backup. Android does not provide app-managed cloud backup of original audio. Restoring it on another device depends on those settings and is not guaranteed. Deleting a meeting removes its local recording on the device processing the deletion. Device voice creates speech using installed system voices. Optional natural narration uses OpenAI-generated voices and caches prepared audio separately for each account on the device; it does not recreate the original speakers. Replaying cached narration sends no new transcript request. Device storage management may remove cached narration, requiring it to be prepared again. Account deletion requests removal of that narration cache on the current device. In the newer recording build, the Live Activity can show a generic recording label, elapsed capture time and a Stop control on supported iPhones with Live Activities enabled, without meeting titles or transcript text. If the control is unavailable, reopen the app to stop. Uploaded transcription audio is removed from our temporary backend storage after each processing attempt. In the newer recording build, completed transcription text and any speaker labels may be retained for up to 24 hours so an interrupted retry can return the result without repeating provider processing or charging cloud minutes again. This temporary result is then deleted, or removed sooner on account deletion; it is separate from meeting transcripts you choose to save. In beta build 14 and later, stopped unfinished recordings remain on the device for retry until saved or discarded; other temporary audio can remain in the app cache. Account deletion includes encrypted cloud journal records and requests cleanup of the journal and recordings on the current device. Other devices may retain offline copies until their local app data is removed. Calendar time blocks you created remain in the selected Apple or Google calendar when a task, app or account is deleted; remove those events separately in your calendar. Disconnecting a task-event link keeps the event. Removing the app removes its local files; keep your recovery code if you use journal sync. Provider logs and backups may follow additional retention schedules, so deletion does not necessarily remove those records immediately.

We use authentication and access controls, but no system can guarantee absolute security.

Personal memory and backups in testing

You can save, correct or forget approved facts and spelling terms, globally or for a selected project. These sync privately with your account through Convex and are cached encrypted on your device when secure storage is available. Relevant approved memory and selected project meeting history can accompany requested AI work. Relevant spelling terms can accompany cloud transcription. Journal entries are not added to this memory automatically. Saved memory is not end-to-end encrypted in the cloud like the optional journal vault.

Memory downloads and unlocked journal backups are readable JSON files. Anyone with access to an exported file can read it; save it somewhere private. Journal restore validates a backup before importing and preserves existing entries, keeping conflicting versions separately. Account exports do not replace an unlocked journal backup or its recovery code. Deleting content from the app cannot delete copies you have exported elsewhere.

Invitations and notifications in testing

You can address an invitation to a verified account email or share a private invitation link through your own messaging app. Recipients review and accept or decline before joining. Links for people who install afterward must be opened again; email-addressed invitations can appear after sign-in with that verified email. Optional remote notifications use Expo and platform delivery services (Apple on iOS and Google on Android) with a device token. In-app notification controls include categories, timing, quiet hours, sound and hidden previews; the invitation inbox remains available without notification permission.

Records that may remain after deletion

Records of voided or refunded purchases used to prevent fraud, safety reports submitted by other people, and moderation review records may be retained for fraud prevention, safety review or legal obligations. These records are not automatically removed by in-app account deletion. Provider security logs and backups may follow separate retention schedules. Contact support for information about retained records or a further deletion request.

Your choices

Edit or delete saved content, revoke permissions, disable AI, and use Settings → Export Data or Settings → Delete account. You may also request access, correction or deletion by contacting us. We may verify ownership; never send passwords or verification codes. See our deletion guide. Cancel subscriptions separately through the store used to purchase them: deleting your account does not cancel Apple or Google Play billing.

Children and changes

This is a general productivity service, not a service directed at children. Contact us if a child has supplied information that should be removed. We update this notice when practices change and show the revision date on this page.