Privacy Policy

Summary

SecureEdge does not collect, transmit, or receive any personal data. The app has no accounts, no analytics, and no third-party software that collects data. The managed build distributed to organizations contains no networking code at all. The App Store build has exactly one network feature, optional model downloads, which never run without your action and which we keep no logs of. Everything you dictate, import, or ask stays on your device, encrypted, and is deleted according to the retention policy your organization's administrator sets. The rest of this document says the same thing in the detail that a privacy policy requires.

Scope

This Privacy Policy describes how Wanyeki Technologies LLC ("we", "us", or the "Company") handles information in connection with the SecureEdge application for iPhone and iPad (the "App") and the website at https://www.secureedgemobile.com (the "Site"). It applies to every version of the App, whether obtained through the App Store, Apple Business Manager, or a mobile device management ("MDM") deployment.

Data collected

The App's App Store privacy declaration is Data Not Collected. The following table states, for each category Apple defines, what the App does.

The App does not use data for tracking, as Apple defines that term, and does not link data to your identity, because it obtains none. In the App Store build, optional model downloads that you start make ordinary HTTPS requests to our model server; we retain no logs of those requests (see Model downloads).

Data that stays on the device

The App processes information so that it can work. All of that processing happens on the device and none of it reaches us. Specifically:

• Dictation. Audio from the microphone is transcribed by Apple's on-device speech recognizer with on-device recognition required. The audio is never written to disk by the App and never leaves the device. The resulting text is cleaned up by the on-device language model.
• Chats and prompts. Your messages and the model's responses are generated on the device and stored in an encrypted local database.
• Documents. PDF, text, RTF, and image files you choose to import are stored encrypted on the device. Text is extracted, and optical character recognition on images is performed, by Apple frameworks on the device.
• Images. Photos, screenshots, or camera captures you attach are analyzed by the on-device model and stored encrypted with the conversation they belong to.
• Audit log. A record of App events (policy loads, model loads, prompts, responses, imports, exports, workflow runs, screenshots, and retention purges) is kept in the encrypted database as a hash chain. It can be exported only when the administrator's policy allows, and only by an action taken on the device.
• Settings. Managed configuration delivered by your MDM is read by the App and never written back.

None of the above is ever transmitted. The managed build of the App contains no networking code and cannot send or receive anything. The App Store build contains networking code for one purpose only, downloading a model at your request, and that code carries none of the data described in this section; it is described in the next section.

Model downloads

The App is distributed in two builds. The managed build, delivered to organizations through Apple Business Manager or MDM, contains no networking code. A build-time test scans every source file for networking symbols, and the compiled binary links no URLSession symbols. Nothing in this section applies to the managed build, which cannot make a network request of any kind. The App Store build, available from the public App Store listing, contains exactly one networking feature: optional model downloads. This section describes that feature.

• Nothing is fetched automatically. The App never makes a network request on its own. The catalog of available models is requested only when you open the download list in the Models screen, and a model file is requested only when you tap Download for that model. If you never do either, the App Store build makes no network request at all.
• Where the request goes. Requests are made over HTTPS to the SecureEdge model catalog, which is served by Amazon CloudFront and Amazon S3 and operated by Wanyeki Technologies LLC, currently in the AWS us-east-2 region. An organization may instead direct the App to its own catalog server through the modelCatalogURL policy key, in which case that server, not ours, receives the request.
• What the request contains. Each request is a standard HTTPS GET. It carries no account, no device identifier, no telemetry, and no content of yours. The only header the App adds is a User-Agent of the form SecureEdge/<version>.
• What we retain. Nothing. Access logging on the model server is disabled, so no per-request records, including IP addresses, are retained by Wanyeki Technologies LLC. As with any HTTPS connection, standard transport data reaches the infrastructure provider (Amazon Web Services) in the course of serving the request, under that provider's own terms.
• Verification. A downloaded model file is hashed with SHA-256 and its manifest signature (Ed25519) is verified before the model can be loaded. If either check fails, the file is deleted.
• Removal. You may remove a downloaded model from the device at any time in the Models screen. Deleting the App removes every downloaded model.
• Background downloads and notifications. A download you start may continue while the App is in the background. If it finishes while the App is in the background, the App posts a local notification on the device. No remote push notification service is involved.
• Administrator control. Organizations may disable downloads entirely with the allowModelDownloads policy key, or point the App at their own catalog with modelCatalogURL. Managed builds ignore both keys because they contain no download code.

Model downloads do not change the App's App Store privacy declaration, which remains Data Not Collected: no data is collected or retained by us in connection with a download.

Device permissions

The App asks iOS for the following permissions, each for one purpose:

• Microphone and Speech Recognition: to dictate. Recognition uses Apple's on-device recognizer only; the App refuses to start dictation if on-device recognition is unavailable for the selected language rather than falling back to a server. Audio never leaves the device.
• Camera and Photo Library: only to analyze images you choose to attach. Images are processed on the device and are not transmitted.
• Notifications: local notifications only, to tell you a background job finished. There is no remote push registration.

An administrator may disable any of these features through policy, in which case the App does not request the corresponding permission. You may revoke a permission at any time in iOS Settings.

Retention and deletion

Chats, transcripts, and imported documents are retained on the device for the period set by your organization's administrator in the MDM policy: the current session only, 24 hours, 30 days, or indefinitely. The default is 30 days. Purges happen automatically and each purge is recorded in the audit log. You may delete individual items sooner. Deleting the App removes everything it stored.

Because we hold no copy of anything, there is nothing for us to delete on request and no retention on our side.

Managed deployments

SecureEdge is designed to be deployed by organizations through MDM. In a managed deployment, your organization's administrator, not Wanyeki Technologies LLC, controls the App's policy: which models may be loaded, how long data is retained, whether dictation, image input, import, export, copy, and custom workflows are available, whether the audit log may be exported, and (in the App Store build) whether models may be downloaded and from which catalog.

If the administrator permits audit export and an export is performed on the device, the exported log is in your organization's possession and subject to your organization's policies, not this one. The audit log records events, timestamps, hashes, and metadata such as character counts, model identifiers, and workflow names; it does not contain the text of your prompts, responses, or documents. Your organization may also apply device-level restrictions (for example, on screenshots) through the MDM. If you use the App on a managed device, consult your organization about its own privacy practices.

Third parties

The App includes no third-party software development kits that collect data. It includes Google's LiteRT-LM inference runtime and Gemma model weights, both of which run entirely on the device and contain no networking. When the App Store build downloads a model at your request, the files are served from Amazon Web Services infrastructure operated by us, as described under Model downloads; no other party is involved in a download. The App does not integrate with any analytics, advertising, crash-reporting, or customer-messaging service. Apple's own frameworks (speech recognition, Vision, PDFKit, CryptoKit) are used on the device under Apple's platform privacy terms.

The Site is a static website. It sets no cookies and includes no analytics or tracking scripts. Our hosting provider records standard web-server access logs (IP address, requested page, user agent, timestamp) for security and operations; those logs are not used to identify individuals and are retained for a limited period.

Children

The App is a business tool intended for adult professionals in organizations. It is not directed to children under 13 (or the applicable age of digital consent in your jurisdiction), and we do not knowingly collect personal information from children. Because the App collects no personal information from anyone, it collects none from children. If you believe a child has used the App in a way that concerns you, contact us at the address below.

No sale or sharing

We do not sell personal information, and we do not share it for cross-context behavioral advertising. We have never done so, and the App's design makes it impossible: no personal information reaches us. This statement is intended to satisfy disclosure requirements under laws such as the California Consumer Privacy Act and comparable statutes.

Security

Data the App stores on the device is protected with iOS Data Protection class Complete, which keeps files inaccessible while the device is locked, and is excluded from device backups. Model bundles are signature- and hash-verified on every load. The audit log is tamper-evident. Details are on the security overview. No method of storage is perfectly secure; the physical security of the device and the strength of its passcode remain with you and your organization.

Your choices

Privacy laws in some jurisdictions grant rights to access, correct, delete, or port personal data, or to object to its processing. We hold no personal data about you, so there is nothing for us to produce, correct, or erase. If you contact us with a request, we will confirm this in writing. Data on your device is under your control and your organization's; you may delete it in the App or by removing the App.

Changes

We may update this Privacy Policy from time to time. The effective date at the top of this page will change when we do, and material changes will be noted on this page. The App does not check for policy updates and cannot notify you in-app; the current policy is always at https://www.secureedgemobile.com/privacy/. Continued use of the App after a change takes effect constitutes acceptance of the updated policy.

Contact

Questions about this policy or our practices: babs@wanyekitech.com.

Wanyeki Technologies LLC
Attention: Privacy