AI Disclosure

Last updated: 17 May 2026. This page explains the AI features that are currently part of Mini Me OS, what data they send to our AI provider, and where the limits are.

The features

Mini Me OS has a small set of optional AI-assisted features. None of them runs automatically.

• Smart Add on Today and Capture. You type a sentence or paragraph and press Smart add. The app sends that text to the AI provider, which proposes a structured list of tasks, goals, projects, routines or notes. You see a preview, edit it, then choose to create the items or cancel.
• Break into steps on the task editor. You press Break into steps. The app sends the task title, the description if you wrote one, and the name of the project the task belongs to (if any). The provider returns 3–6 suggested subtasks. You tick which ones to keep, then create them.
• Break a project into tasks on the project view. You press the button. The app sends the project title (and optional description) and the provider returns 5–10 concrete first steps. You pick which ones become real tasks before anything is saved.
• Explain today's plan on the Assistant page. You press the preset. The app sends a compact snapshot of Today: how many tasks are on, the title of the top task and the reason it ranks highest, the titles of tasks that can wait, a short list of attention items (overdue, stuck projects, goals with no work, capture backlog) and the workload status. The provider returns a 60–120 word summary in plain English.
• Voice → text fallback on Capture (Pro and Max). Chrome / Edge / Safari / Opera run the browser's own on-device speech engine for free and nothing leaves your device. On Firefox and in-app webviews — where on-device speech isn't available — Pro and Max can send a short audio clip to the AI provider for transcription. The transcript drops into the textarea so you can review and edit before Smart Add does its thing.

What is not sent

• Your email address.
• Your password (we never have it in plaintext anyway — Supabase Auth holds the hash).
• Your full task list. Each feature only sends the text it explicitly needs.
• Your goals, routines, reviews, decisions or capture history beyond what is specifically required for the feature you triggered.
• Data from other users. The AI calls are scoped to the request you made and never include another account's content.

AI provider

The AI features are served by Google (Gemini API). The operator chose Google (Gemini API) and configured an API key on the server. Google (Gemini API) processes API requests under their own data-usage terms; by default they do not train shared models on customer API content. If the operator removes the API key, all three features automatically fall back to a built-in rule-based version that runs entirely on our server and sends nothing to a third party.

How the output is treated

• AI output is always shown as a preview first.
• Mini Me OS only writes data to your account after you click the explicit confirm button (e.g. "Create these", "Create selected tasks").
• You can cancel the preview without saving anything.
• The previews can be edited before you save them.

Logging

We do not log AI prompts or responses to our own servers. Google (Gemini API) logs API requests on their side according to their own retention policy. We do not have a copy.

Limits and honest expectations

• AI suggestions can be wrong. Read the preview before you save it.
• The model has no memory of your past data. Each call sees only the text you submitted for that call.
• The model does not access your contacts, files, social accounts or any third-party system. The two opt-in integrations Mini Me OS has — Calendar Sync (read-only iCal) and the Email module (read-only Gmail on Max-tier) — are isolated from the AI calls: their contents are never sent to the AI provider.
• If the AI provider is slow, unavailable, or the API key is missing, the rule-based fallback kicks in for the text features. The button always produces a result; it just may be less clever.

Turning AI off

There is no on/off toggle for AI today, but every AI feature only runs when you press its button. If you never press Smart add, Break into steps, Break a project into tasks, Explain today's plan, or use the voice button in a browser without on-device speech, no text or audio leaves the app for AI processing. The regular add buttons (Add, Save) never call AI.

Related

For the broader privacy picture see the Privacy Policy. For the role of Google (Gemini API) as a sub-processor see the Data Processing Information.