AI data, privacy and where it runs

Vivotics AI is built so you always know what it can see and where your data is processed. This page explains both, so you can use it with confidence on client work.

What the AI can see

Every AI feature runs under the permissions of the person using it. The assistant only ever reads records that user can already open in the app: their tasks, projects, docs and meetings. It cannot surface anything a user is not allowed to see, and private tasks stay private to their owner and assignees. Answers cite the exact records they came from, so you can check the source.

Where your data is processed

The active AI provider, its processing region and its data-use terms are always shown under Settings → AI, in the "Where your AI data is processed" panel, so an administrator can verify where content goes before the team relies on it. On our cloud the models can run on Vivotics' own servers, so workspace content does not have to leave our stack at all, and whichever provider is active, your inputs are not used to train shared models.

The AI provider is pluggable. A self-hosted or self-managed deployment can point Vivotics at a different provider (for example an OpenAI-compatible endpoint, a regional host, or a model running on your own infrastructure). Settings → AI always shows the current provider and flags any endpoint that would send data outside a trusted region, so an administrator can review it before AI is used on sensitive data.

Meeting transcripts

The AI Notetaker turns meeting audio into a transcript and then into minutes and tasks. Transcription can run on a self-hosted endpoint so the audio never leaves your stack, and the text steps use the same trusted AI provider as the rest of the suite.

Metering and control

• AI is a paid add-on unlocked per person with an AI seat; only seat holders can run it.
• All usage is metered against a shared monthly pool, visible in Settings → AI.
• Administrators control seats, the team glossary, and the AI provider.

In short

• The AI sees only what each user already can.
• The active provider and region are shown in Settings → AI, and inputs are not used to train shared models.
• The provider and region are always visible in Settings → AI, and a non-trusted endpoint is flagged before use.