Bimbi Privacy Policy

Effective date: September 25, 2026

This Privacy Policy explains how Bimbi’s creator, operating the project under the name Bimbi Labs (“Bimbi Labs,” “Bimbi,” “we,” “us,” or “our”), collects, uses, discloses, and protects personal information when you use getbimbi.com, the Bimbi iOS app, the Bimbi web beta, Nana, our former website demo, and related support and beta services (collectively, the “Services”).

The short version: Your family’s information is used to provide Bimbi, never for marketing or sale. In supported iOS versions, accepting the core service lets Bimbi keep your baby’s profile, setup answers, and sleep and care records available across your devices. You can still track offline. Nana sends conversations and relevant family context to OpenAI after your permission. You can delete your family data; optional analytics, Shared learning, and model training are not included in core service acceptance.

1. Who we are

Bimbi is an independent project operated by its creator under the project name Bimbi Labs. It provides voice-first parenting tracking, education, and coaching for adult parents and caregivers. Bimbi is not directed to children.

For privacy questions or requests, contact our Privacy Lead at privacy@getlullay.com. General support is available at hello@getlullay.com, and safety concerns can be sent to safety@getlullay.com.

2. Information we collect

The information we collect depends on the Services and app version you use. A supported iOS version explains synchronization of selected baby profiles, setup answers, and sleep and care records before you accept the core service. Existing profiles and history are uploaded only after that acceptance and any necessary profile linking. An app update does not silently expand an earlier permission. Earlier versions may offer separate optional sharing controls; those choices remain valid only for their original purpose.

Account and contact information

We may collect:

• your Sign in with Apple identifier and, if Apple provides it, your name or email address, including an Apple private-relay address;
• for the limited web beta, your name, email address, password hash, language, consent timestamp, and session information;
• language, time zone, account status, trial status, and preferences;
• waitlist information, support messages, privacy requests, safety reports, and beta feedback; and
• any contact information or attachments you choose to include when contacting us.

We do not receive your Apple Account password.

Child, caregiving, and tracking information

You may choose to provide information about a child and your family’s routines, including:

• the child’s name, date of birth, age, prematurity or adjusted-age information, and whether solids have started;
• setup answers, including what you prefer to be called, birth timing, feeding, usual waking and bedtime ranges, naps, settling methods, night wakings, your priorities and preferred support, and healthcare instructions you choose to share;
• optional reported sex, health status, and coded known conditions, including whether you report a clinician diagnosis; these are caregiver reports, not verified medical records;
• sleep, nap, waking, feeding, mood, schedule, bedtime, plan, and routine information;
• sleep setting, sleep challenges, caregiver responses, goals, method preferences, and comfort with different approaches;
• notes, summaries, plan progress, guides, and facts you ask Nana to remember; and
• information you type, say, or confirm while using Nana or tracking features.

This information can reveal health, development, and family-routine details. We treat it as sensitive personal information and, where applicable, consumer health data.

Conversations, voice, and AI information

When you start a Nana voice or chat session, the Services may process:

• microphone audio and speech transcriptions;
• typed messages and Nana’s responses;
• the child’s name, age, prematurity, feeding and sleep context, recent tracking history, active plan, language, time zone, summaries, and relevant facts; and
• tool actions needed to log an event, update a plan, create a guide, or complete another request you make.

In the iOS app, raw voice audio is streamed for real-time processing and is not stored by Bimbi Labs. Partial captions are discarded. Finalized conversation text is stored on your iPhone only if you enable conversation history; summaries may be stored locally so you can review prior sessions. The Bimbi iOS backend is designed not to store raw audio, typed chat content, transcripts, or summaries.

The separate web beta stores typed messages and finalized voice transcript lines, summaries, guides, and plan information on Bimbi’s systems so the beta experience can persist between browser sessions. The web beta never stores raw audio.

Optional Shared learning in supported iOS versions

Shared learning lets an adult caregiver explicitly review and contribute structured observations about a completed coaching outcome. Contributions can include:

• age in months, adjusted age, weeks born early, reported sex or a not-shared value, and an optional coded health report frozen when the outcome was evaluated;
• a profile revision, report and evaluation dates, and whether a condition was parent-reported or reported as a clinician diagnosis;
• the method and goal codes, an outcome such as improved, unchanged, worse, or insufficient data, aggregate baseline and follow-up sleep measures, adherence, coded factors that may affect the observation, and the evaluation policy/version; and
• account-linked pseudonymous child and outcome identifiers used to avoid duplicates, apply corrections, and honor deletion.

This submission excludes names, dates of birth, private notes, free text, raw sleep logs, audio, and conversation transcripts. The optional structured health fields include asthma, eczema, reflux, food allergy, sleep apnea, or an “other” code without a diagnosis description. You may leave sex or health information unshared. Contributions remain linked to your Bimbi account; hashing identifiers does not make this data anonymous.

Changing a profile does not silently add today’s health information to an older outcome. Editing an outcome invalidates its older contribution when the app can reach Bimbi; you must explicitly review and submit a corrected outcome to share it again. There is no automatic retrospective upload.

Purchase and subscription information

Apple processes App Store purchases. We may receive product identifiers, transaction and original-transaction identifiers, subscription state, expiration or revocation status, storefront or environment information, and an app-account token used to associate an entitlement with your Bimbi account. We do not receive or store your full payment-card number.

Device, usage, and operational information

We may collect limited technical and operational information, including:

• app version, build number, operating-system version, device model, language, time zone, and timestamps;
• a pseudonymous or hashed account identifier;
• session start and end times, model tier, duration, token counts, estimated or measured service cost, trial usage, and rate-limit counters;
• content-free product events, counts, timings, and short categorical values if you opt in to app analytics; and
• IP address, browser type, security signals, essential session cookies, and approximate region inferred from IP when you use the website, web beta, waitlist, or demo.

Optional in-app analytics contain account-linked counts, timings and short feature categories, not message text, transcripts or private notes. They are not anonymous. Essential service usage and security records are separate from optional analytics. The private operations dashboard displays usage, reliability, cost, subscriptions, consented product events, and feedback and optional reply email you submit. Authorized personnel may access synchronized records and contributions you explicitly share only as needed to operate the applicable service, address a support or safety issue, or honor correction and deletion. Core synchronization does not authorize unrestricted internal access or use for research or model training.

Information we do not currently collect

The current native app does not request access to contacts, HealthKit, the photo library, camera, Bluetooth data, precise device location, advertising identifiers, voiceprints, or biometric identifiers. This does not exclude attachments you deliberately send in a support request outside the app. Nana does not perform cry detection, speaker identification, emotion recognition, or voiceprint analysis. A child’s voice or other background sound may be incidentally captured during a Nana voice session; avoid starting a session when others should not be recorded.

3. Where information is stored

Native iOS app

The app keeps a local copy of your records on the iPhone so tracking can work offline. In versions offering cloud synchronization, accepting the core service also stores the baby profiles, setup answers, and sleep and care records you choose to synchronize, including existing entries, corrections, timestamps, and tracking notes, on Bimbi’s systems. Signing in with the same account on another supported device lets you access those synchronized records. Separate caregivers are not automatically linked to the same child.

Bimbi uses DigitalOcean for the synchronized profile and tracking database and Cloudflare for the application backend, authentication records, operational records, and access controls. Each account can contain separate profiles for more than one child. Account and child identifiers keep each child’s records connected to the correct account; these linked identifiers are personal information, not anonymous data.

Local plans, evaluations, optional learning profiles, Nana summaries, saved conversation history, and private profile notes are not part of this version’s cross-device synchronization. They may still be included in the relevant context sent to Nana with your AI permission. Depending on your Apple settings, local app data may be included in an Apple-controlled device backup.

The app also sends information needed to authenticate, verify subscriptions, operate Nana, submit feedback, or carry out optional sharing you separately authorized. Earlier Shared learning contributions remain in an account-linked Cloudflare database and can be withdrawn.

Web beta

The web beta stores account, child profile, tracking, plan, guide, conversation, consent, and product-event information in Bimbi’s Cloudflare-hosted database. This architecture is separate from the native iOS app.

Website and former demo

The public Nana demo was discontinued on September 19, 2026. Visitors can request access to use Nana in the app. The former demo did not require an account, and Bimbi did not store its audio or transcripts. Existing transient security records remain subject to the retention terms below. Waitlist submissions are stored so we can contact you about Bimbi.

4. How we use information

We use information to:

• create and secure accounts and authenticate sign-ins;
• provide tracking, synchronize accepted profiles and sleep/care records across your devices, and provide schedules, reminders, summaries, plans, guides, and Nana conversations;
• personalize the Services using the child and family context available to the selected product;
• process subscriptions, trials, restores, and entitlements;
• respond to support, privacy, safety, and beta-feedback requests;
• prevent abuse, fraud, and unauthorized access and enforce usage limits;
• measure reliability, service cost, and product performance;
• troubleshoot and improve the Services using content-free analytics and feedback you intentionally submit;
• where you separately choose Shared learning, examine patterns in reported profiles and coaching outcomes to improve the product and evaluate future changes;
• comply with law, enforce our Terms, and protect users, children, Bimbi Labs, and others; and
• communicate material service, safety, legal, or account changes.

We never use child information or the personal information you provide through the native app and Nana for marketing, advertising, or sale. We do not add native-app accounts or family information to marketing audiences. A separate website waitlist or newsletter submission is used only to send the communications you requested through that form; it is not combined with family or coaching records. We do not use customer content to train or fine-tune our own or a third party’s general-purpose model unless we first offer a separate, explicit opt-in. Shared learning does not grant permission to train or fine-tune a general-purpose model. It is observational product learning, not clinical research, proof that a method caused an improvement, diagnosis, or automatic treatment. There is no automatic model training from these contributions. Small-group patterns can be misleading; aggregate reports suppress groups with fewer than ten distinct children and ten accounts.

5. Service acceptance and permissions

In supported iOS versions, one core-service acceptance covers the baby profiles and setup answers you choose to synchronize, together with their existing and future sleep and care records. The app asks you to identify an existing baby when linking is needed; matching a name or birth date alone does not give someone access. Earlier profile-only or local-only permissions are not silently expanded to include history.

Nana has a brief, separate permission identifying OpenAI and the conversation and family context used for coaching. You can use tracking without enabling Nana. Turning Nana off in Settings stops future AI requests; it does not undo processing that already occurred or delete saved records. Microphone and Apple speech permissions remain under your control.

Core-service acceptance and Nana permission do not enable optional usage statistics, Shared learning, saved conversation history, research, or model training. New beta users are not enrolled in optional analytics or Shared learning. If an earlier version let you enable these features, you can still withdraw that permission and request removal of the associated records in Settings. Earlier refusals and withdrawals remain respected.

Saved conversation history is off by default. Turning it off stops future local transcript copies; it does not erase copies already saved or alter an AI provider’s retention. Separately provided health details are used only within their disclosed scope; private health notes are excluded from structured AI context.

Where consent is the legal basis for processing, you can also withdraw it for future processing by contacting privacy@getlullay.com. Withdrawal may make the related feature unavailable.

6. AI and voice processing

Nana uses OpenAI’s API services. During a Nana session, microphone audio may travel directly from the device to OpenAI over an encrypted real-time connection, while Bimbi’s backend handles authentication, safety instructions, selected context, tools, and connection signalling. Typed conversations and recap or fact-extraction requests pass through Bimbi’s backend to OpenAI. Knowledge searches may send the search text to OpenAI to create a search embedding. Optional Apple speech dictation is a separate feature: recognition runs on the iPhone when supported; otherwise Apple processes the dictation audio under its speech-service terms and permission prompt.

OpenAI states that API customer content is not used to train its models unless the API customer opts in. Under OpenAI’s default API controls, abuse-monitoring logs may contain prompts, responses, audio, transcriptions, and related metadata and may be retained for up to 30 days, unless law requires longer retention. Bimbi Labs has not authorized OpenAI to train models on Bimbi customer content. We do not promise zero provider retention. See OpenAI’s API data controls for endpoint-specific policies.

Bimbi personnel do not routinely listen to or read native iOS Nana conversations because Bimbi does not store them. Access to web-beta conversation content, support messages, safety reports, or feedback is limited to authorized personnel who need it to provide support, investigate a reported problem, protect safety or security, or comply with law.

AI can be inaccurate or incomplete. Nana is an educational and coaching feature, not a healthcare professional or emergency service. See our Safety page and Terms of Service.

7. When we disclose information

We disclose personal information only as described here:

• OpenAI: real-time voice, transcription, chat, recap, fact extraction, and related AI processing;
• Cloudflare: website, web-beta, backend, authentication and operational databases, security, rate limiting, and content delivery;
• DigitalOcean: hosting and backups for synchronized baby profiles and sleep/care records;
• Apple: Sign in with Apple, App Store subscriptions, optional speech recognition, notifications, device services, and any Apple-controlled backups;
• Plausible: cookieless website analytics configured without advertising profiles;
• Support and communications providers: email delivery and the tools we use to answer messages; and
• Professional advisers and authorities: where reasonably necessary for legal, security, accounting, insurance, safety, or compliance purposes.

We may disclose information when you direct us to, when required by valid legal process, or when reasonably necessary to protect a person from serious harm, investigate fraud or abuse, or protect legal rights. If Bimbi Labs is involved in a financing, merger, acquisition, reorganization, or sale of assets, information may be transferred subject to confidentiality and applicable law.

We do not sell personal information, share it for cross-context behavioural advertising, or allow third-party advertising trackers in the native app. We do not knowingly sell or share consumer health data.

8. Cookies and website analytics

The public website uses Plausible for aggregate, cookieless analytics and Cloudflare for security. The web beta uses an essential secure session cookie to keep you signed in. Cloudflare Turnstile may process technical and security signals when you use protected forms. We do not use advertising cookies or respond differently to browser “Do Not Track” signals because we do not track users across unrelated services for advertising.

9. Retention and deletion

While your account is active, we keep useful profile, setup, and tracking history for as long as it is needed to provide the services you use and within your applicable permissions. We do not impose a routine expiry on that useful history. You can correct or delete it. Account deletion removes personal and account- or baby-linked product and learning records from Bimbi’s active systems, including synchronized profiles and setup answers, sleep/care history and their saved corrections, account records, and submitted contributions. Deleting one baby removes that baby’s records while preserving other profiles.

• Your devices: local deletion happens on the device where you request it. Other devices need to reconnect for synchronized deletion to reach them. Plans and conversation history that are not synchronized must be removed on each device. Apple-controlled backups and copies you export are managed separately.
• Pending deletion: server deletion needs an internet connection. If deletion is interrupted, the app keeps the request pending for retry and does not report completion. Records that have not yet been erased remain until the relevant cleanup succeeds. If account-record removal succeeds but final technical-marker cleanup is interrupted, Bimbi retries that remaining cleanup automatically. The remaining marker contains only the identifiers and completion details needed for this purpose; it is not used for learning or any other purpose.
• Deletion safeguards: while your account exists, minimal records of deleted children and tracking entries prevent an old device or delayed request from bringing them back. These records contain identifiers and deletion details, not names, dates of birth, notes, conversation content, or tracking history. They remain linked to your account and are used only to enforce deletion. Successfully completed account deletion removes these records and the account’s own technical deletion marker from Bimbi’s active systems. No account-level deletion marker remains after successful account deletion.
• Optional analytics and Shared learning: withdrawal deletes associated active event or contribution records. Minimal consent and retraction records prevent delayed requests from reversing a withdrawal while the account exists; account deletion removes those records. Offline requests remain pending until acknowledged.
• Database backups: DigitalOcean’s managed PostgreSQL backups are retained for seven days. Cloudflare D1 database recovery history lasts seven days on its Free plan or 30 days on its Paid plan. Deleted records may remain in these restricted backups until they expire. Backups are for recovery, not product learning or normal access; any recovery must preserve subsequent deletions and withdrawals before service resumes.
• AI processing: OpenAI’s default API abuse-monitoring retention is separate from Bimbi’s database backups and may retain eligible content for up to 30 days, unless law requires longer. See section 6.
• Web beta, website, and support: the separate web-beta account has its own deletion flow. Website waitlist and support records remain only as needed for the request or relationship, or a legal obligation. Contact us to delete records supplied outside your native app account.
• Purchases and legal obligations: Apple manages its purchase records. Any retention required by law is limited to the records and period required; it does not permit keeping your deleted family history for learning.

Only aggregate insights or model improvements that cannot reasonably be linked back to you or your family may remain after deletion. A pseudonymous identifier or a small group count is not anonymization. This is a limit on what may be retained, not permission to use your records for a new purpose or a claim that Bimbi currently trains models from family data.

Deleting Bimbi data does not cancel an App Store subscription. Cancel it separately in your Apple subscription settings. Apple-controlled device backups, provider records, and exported copies follow their applicable controls and retention schedules.

10. Your privacy rights and choices

Depending on where you live, you may have rights to:

• know whether we process your personal information and obtain access to it;
• request a portable copy;
• correct inaccurate information;
• delete information;
• withdraw consent or object to certain processing;
• opt out of sale, targeted advertising, or certain sharing—although Bimbi does not engage in those practices;
• appeal a denied privacy request; and
• make a complaint to the privacy or data-protection authority where you live.

Use Settings → Privacy & data → Manage data in the current iOS beta to export the records available on that device, delete a baby, or delete your account. Earlier versions may place these controls under More or Privacy center. The export states its coverage; it is not a complete server backup. The separate web beta has its own controls. You can also email privacy@getlullay.com. We may need to verify your identity and caregiver authority. An authorized agent may submit a request where permitted by law, subject to appropriate proof. We will not discriminate against you for exercising a privacy right.

11. Children and caregiver authority

The Services are for adults. You must be at least 18 or the age of legal majority where you live. Children may not create an account or intentionally use Nana.

By providing information about a child, you confirm that you are the child’s parent or legal guardian or are otherwise authorized by the parent or guardian to provide the information and use the Services for that child. Babysitters and other caregivers must have the parent’s or guardian’s permission.

We do not knowingly solicit personal information directly from children. If you believe a child has created an account, intentionally spoken to Nana without adult involvement, or provided personal information without appropriate authorization, contact privacy@getlullay.com. We will investigate and delete information where required.

12. Security

We use administrative, technical, and organizational safeguards designed to protect personal information, including encrypted network connections, Apple platform protections, access controls, pseudonymous account identifiers in the native backend, and data-minimization rules for analytics. No service can guarantee absolute security. Keep your device, Apple account, web-beta credentials, and email account secure, and contact us if you suspect unauthorized access.

13. International processing

The synchronized profile and tracking database currently uses DigitalOcean in Toronto, Canada. This does not mean all information stays in Canada: Bimbi Labs and its other providers, including Cloudflare and OpenAI, may process information in Canada, the United States, and other countries where they operate. Those countries may have privacy laws different from the laws where you live. Where required, we use contractual and other safeguards intended to protect information transferred across borders.

14. Third-party services and links

The Services may link to Apple, healthcare, crisis, educational, or other third-party resources. Their privacy practices apply to information you provide directly to them. This Policy does not govern third-party services that Bimbi Labs does not control.

15. Changes to this Policy

We may update this Policy as the Services, vendors, or law change. We will post the revised Policy with a new effective date. If a change materially affects how we use sensitive information, we will provide additional notice in the app, by email, or through another reasonable channel before the change takes effect where required.

16. Contact us

Questions, requests, or complaints may be sent to:

Bimbi Labs — Privacy Lead

privacy@getlullay.com

For general support: hello@getlullay.com

For urgent product-safety concerns: safety@getlullay.com