Data Use

Last updated: 2026-05-15

This page describes how Minima Technologies, Inc. ("Minima") uses your content. It supplements the Terms of Service and the Privacy Policy.

By default, Minima does not use your content to train AI models. Your sketches, files, prompts, and conversations are used to operate the product — to render audio and video, to run agents, to sync across your devices, and to recover your work when something goes wrong.

We do plan to use customer content to improve our AI models, but only with your explicit, opt-in permission. This page describes what we use today, what changes if you opt in, and the limits on what consent can do once a model has been trained.

We use the data you give us — and the data we generate as you use Minima — to run the service you are paying for. That includes:

• Sync and persistence. Your projects, files, folders, and settings are stored so you can pick up where you left off, on any device.
• Rendering. Audio, image, and video pipelines process your sketches into deliverable artifacts.
• Agent and AI features. Prompts and the relevant context are sent to AI providers to produce a response, and the result is returned to you.
• Billing and usage accounting. Credit balances, invoices, and rate limits are computed from usage data.
• Security and abuse prevention. Logs and rate-limit counters help us detect abuse and respond to incidents.

None of these uses involve training AI models on your content. Training is a separate activity governed by the section below.

Minima orchestrates several large language models. When you use an AI feature, the prompt and the necessary context are forwarded to the model provider you have selected (or that we route to).

• Anthropic (Claude). Used for chat, agent runs, and reasoning. Anthropic does not train on commercial API traffic by default.
• OpenAI. Used as a fallback and for specific capabilities. OpenAI does not train on API traffic by default.
• Google (Gemini). Used for selected features and for OAuth sign-in. Google does not train on API traffic by default.

We send the minimum prompt and context required for the request. We strip identifying details from system prompts where we can, and we scope per-user context so that one customer's data is never embedded in another customer's prompt.

To make your work searchable and to ground AI responses, Minima generates vector embeddings of your content. These embeddings live in a Chroma collection scoped to your account or organization, and they are never shared across customers.

Embeddings follow the same lifecycle as the source content. When you delete a file, project, or conversation, the corresponding embeddings are deleted too. Because embeddings are derived from your content, they can always be regenerated; they are not a separate canonical store.

Minima keeps your AI conversations so you can return to them, fork them, and build on them. We try to keep only what is useful.

• Conversation history is retained for 12 months by default, and you can shorten or extend that in settings.
• Intermediate "thinking" traces — the model's scratch work — are retained for 30 days by default. They are useful for debugging and rarely useful long-term.
• "Forget this conversation" deletes both the conversation rows and the embeddings derived from them.

When Minima renders audio, image, or video that was produced or meaningfully shaped by AI, we mark the output as artificially generated using C2PA-style provenance metadata. This helps your collaborators, distributors, and the broader internet understand what they are receiving.

The chat surface clearly identifies that you are talking to an AI, consistent with Article 50 of the EU AI Act and California's bot disclosure rules.

Default: off. We do not use your content to train AI models unless you opt in. New accounts start with training consent disabled, and there is no setting that enables it automatically.

If you opt in, we may use selected content from your account to fine-tune or evaluate AI models we operate, in order to make Minima better at the kinds of work you do. You can review and toggle this in your account settings, and you can change your mind at any time.

What "selected content" means:

• Included if you opt in: sketches, code, node graphs, knob states, prompts, and AI conversations from projects you have not explicitly excluded.
• Always excluded: billing and payment data, invoices, audit and security logs, and account credentials.
• Always excluded: profile information from OAuth providers (your Google, Apple, or GitHub identity).

Revocation. You can revoke training consent at any time. From the moment you revoke, we will stop using your new content for training and we will remove your content from training datasets that have not yet been used.

The honest limit on revocation. If your content has already been used to train a model, that training cannot be undone — model weights are statistical and cannot be edited to remove a single contributor. Revocation prevents future use; it cannot retroactively strip your data from a model that has already learned from it. We tell you this clearly so consent is meaningful.

We do not sell your content. Training, when you opt in, is performed by Minima for Minima's own models. We do not sell, license, or share your content with third parties to train their models.

Every account has a privacy panel that gives you direct control over how your data is used.

• Export a machine-readable archive of your files, folders, settings, conversations, and credit history.
• Delete your account, which cascades through our primary database, local mirrors, vector embeddings, email contact, and analytics.
• Pause AI for your account if you want to use Minima without sending anything to AI providers.
• "Forget this conversation" on any individual AI conversation.
• Toggle training consent on or off at any time.

If we ever expand how we use your data — for training or anything else — we will update this page, notify you in advance, and re-prompt for consent. We will not silently change the default and we will not opt you in retroactively.