Privacy policy

1. Who we are and what this policy covers

Noice Pty Ltd (ABN 91 634 536 360) operates Noice.work, a hosted platform for creating, managing and publishing websites, content and media. This policy explains how Noice handles personal information when people visit Noice.work, use an account, work in a customer site, submit a form on a site, or contact us about the service.

We handle personal information in line with the Privacy Act 1988 (Cth) and the Australian Privacy Principles. A customer may also have its own privacy obligations for the content and forms it runs through Noice.work.

2. Information we collect

The information we collect depends on how you use Noice.work. It can include:

• Account information: name, email address, role or job title, profile details, avatar, sign-in identity, organisation membership and account-security metadata.
• Billing information: plan, order and invoice details, billing contact and payment status. Payment-card details are provided to our payment processor directly and are not stored by Noice.
• Service activity: sessions, audit and workflow activity, notification preferences, API-key names, assistant conversations, prompts, approvals and other actions taken in the service.
• Customer content: pages, documents, brand settings, uploaded files and generated image, video, audio or other media that a customer or its users choose to store or process.
• Connected-service information: when an organisation connects an Instagram account, Meta can provide the account or Facebook Page identifier, display name, granted permissions, access token and token expiry. Noice.work also processes the finished media and caption that an authorised user approves for publishing.
• Form submissions: the values a visitor chooses to submit through a form on a customer site, which may include contact details and other personal information requested by that customer.
• Site search terms: the normalized words a visitor enters into public site search and the number of results returned. Noice.work aggregates this by customer site, UTC day and term without a visitor or session identifier. Customer site administrators and owners can read these terms and counts. A daily count can still be one, and the term can contain personal information.
• Usage and technical information: consent state, a first-party visitor identifier after consent, pages requested, audience assignment, browser or request information used to deliver and secure the service, and service errors.

Please do not provide personal information that you are not authorised to use or ask us to process.

3. Why we use personal information

We use personal information to provide and secure Noice.work, authenticate users, keep customer sites separated, publish and deliver customer content, run requested AI and media features, process form submissions, provide support, administer accounts and billing, investigate faults or misuse, and meet legal obligations.

Depending on the circumstances, we rely on performing our agreement with a customer, steps requested before entering that agreement, consent, compliance with law, and our legitimate business purposes in operating and protecting the service. Visitor-level analytics and audience personalisation are consent-gated as described below. Identifier-free operational aggregates, including daily site-search totals, do not use an analytics cookie or visitor identifier.

4. Service providers and subprocessors

Noice.work uses external services only where their feature is configured. The product registry names the following providers:

• OpenRouter: AI model routing for content composition, editing, assistant and document-understanding features. Requests can include prompts and the customer content needed to perform the requested task, and are routed to the model host configured for the feature, such as DeepSeek, Google or Moonshot AI.
• Amazon Web Services: transactional email (Amazon SES) and document text extraction (Amazon Textract), both in the Sydney region. Email carries the recipient address and message content; extraction receives the uploaded document being processed.
• Stripe: payment processing for paid plans. Stripe receives billing contact and transaction details; payment-card details are provided to Stripe directly and are not stored by Noice.
• Google and Microsoft: sign-in. When you sign in with Google, or with Microsoft Entra where an organisation has configured it, the provider shares your email address and sign-in identifier with Noice.work.
• Meta (Instagram and Facebook): optional Instagram account connection and content publishing. When an organisation administrator connects an account, Meta shares the connected account or Page identity, permissions and tokens needed to keep the connection working. Noice.work sends Meta only the media, caption and publishing request that an authorised user approves; Meta then hosts and distributes the resulting Instagram post under its own terms and privacy policy.
• PostHog: consent-gated visitor analytics and audience personalisation. It can receive a first-party visitor identifier, page activity and audience properties only after analytics consent. A customer site sends analytics only to the PostHog project its organisation connects, or nowhere at all.
• ElevenLabs: optional voice and narration generation. Depending on the selected feature, requests can include live microphone audio and spoken intent, or written copy for narration.
• fal: optional image, video, 3D and music generation. Requests can include the prompt and any reference media the customer supplies for the selected feature, and for music a bounded prompt and duration.
• Meshy: optional 3D model generation. Requests can include the prompt and any reference imagery the customer supplies.

Cloudflare provides the compute, database, object-storage, vector-search, image and edge-delivery infrastructure used by Noice.work. The content library that grounds search and assistant responses is part of Noice.work itself, stored on that Cloudflare infrastructure, not a separate provider. If a configured provider is unavailable, the related feature may be unavailable or fall back to a non-personalised experience. We do not silently send a request to a different deployment.

5. International transfers

Noice is based in Australia, and Noice.work runs on Cloudflare's global edge network. Transactional email and document text extraction run in Amazon Web Services' Sydney region. AI model routing through OpenRouter and the platform's own analytics are United States based, and the payment, sign-in, social-publishing, voice and media-generation providers named above may process information outside Australia depending on the provider and feature used. Where Australian Privacy Principle 8 applies, we take reasonable steps in relation to overseas disclosure. Customers should assess whether optional AI, social-publishing, voice or media features are suitable for the information they choose to submit.

6. Cookies, analytics and personalisation

Noice.work uses essential cookies for sign-in, security and remembering choices. Customer sites may also ask for first-party analytics consent.

PostHog feature flags and personalisation are evaluated only after a visitor accepts analytics consent. Before consent, or after consent is declined, the site serves the default visitor experience without a PostHog audience decision. Declining consent clears the first-party visitor, audience and experiment cookies. A customer site may provide more information about its own forms, content and cookie choices.

Public site-search totals are separate from PostHog visitor analytics. They do not use an analytics cookie or visitor identifier and are collected as first-party, site-level operational aggregates even when PostHog consent is absent or declined.

7. Security

We use technical and organisational safeguards designed to protect information, including authentication, role and permission checks, tenant-scoped storage, controlled administrative actions and audit records. No internet service can guarantee absolute security. Please contact us promptly if you believe an account or customer site has been compromised.

8. Retention

We keep information for as long as needed to provide the service, meet legal or accounting obligations, resolve disputes, protect the service and apply the retention settings or deletion actions available for the relevant customer site. Retention varies by the type of record and the customer's configuration, so this policy does not promise a single period.

An active Instagram connection keeps its encrypted provider token until it expires, is replaced or the organisation disconnects it. Disconnecting immediately removes the credential bytes while retaining bounded non-secret connection and audit metadata. A valid Meta data-deletion request removes the matching connection record and scrubs its provider subject from retained audit detail.

Some security and publishing records retain an opaque internal actor identifier. If the account record that links that identifier to a person is erased, those records become pseudonymised. Customer-controlled page versions and free-text content may also be governed by the customer's content-retention choices.

Before aggregation, a search delivery message may remain in a Cloudflare Queue for up to 24 hours and is not copied to a dead-letter queue. Public site-search terms are then retained in the live service as daily site-level aggregates for 90 days by default. A customer site administrator can choose another bounded period, up to a maximum of 365 days, or set the period to 0 to stop collection and clear live totals. The aggregate stores no visitor or session identifier, but a term may contain personal information. Historical service-backup copies, when present, can outlive the live row and currently have no configured age-out or selective re-erasure window.

9. Access, correction, export and erasure

You can ask to access or correct personal information about you, obtain an export, or request erasure. Email hello@noice.net.au. We may need to verify your identity and confirm the organisation or customer site involved before acting.

Noice.work can collect an access report or export for an identified member or email address. Erasure removes personal information held in structured, identified fields, including the user record and related sign-in records, sessions, account credentials, assistant conversations, avatar bytes, workflow name or email labels, and form submissions keyed to the exact email address.

Audit and page-version attribution can retain only the opaque internal actor identifier after the identifying account record is removed. Free-text fields are handled on a best-effort basis and through retention processes; they are not searched for every unstructured mention of an email address because that could erase another person's data. Erasure of a consent-gated PostHog visitor profile is a manual step in the request process.

Public site-search aggregates have no subject identifier, so Noice.work cannot identify or selectively erase one visitor's rows from that store. Live rows are removed through the customer site's retention policy, and disabling collection clears the site's live search totals.

10. Privacy complaints

Please send a privacy question or complaint to hello@noice.net.au so we can investigate it. If you are not satisfied with our response, you may complain to the Office of the Australian Information Commissioner.

11. Changes to this policy

We may update this policy when the service, our providers or legal requirements change. We will update the date at the top of this page when we do.

12. Contact us

Noice Pty Ltd
Level 6, 343 Little Collins Street
Melbourne VIC 3000, Australia

Email: hello@noice.net.au
Phone: 1300 572 008
ABN: 91 634 536 360