Does Jarvis Train Its AI On My Data?
Does Jarvis (getjarvis.eu) Train Its AI On My Data?
No. Jarvis (getjarvis.eu) never trains any AI model on your data. Your prompts, the screenshots it reads, the persistent memory it keeps, and anything pulled from connectors like Gmail, Slack, Notion, or GitHub are never used to train or fine-tune a model, neither Jarvis's own systems nor the upstream providers it routes to. Jarvis sends tasks to frontier models from Anthropic, OpenAI, and Google under contractual no-training terms, so Anthropic, OpenAI, and Google do not absorb your content into their datasets either. This holds across macOS, Windows, and Linux. Jarvis runs as a floating bar you summon with Cmd+/ or Ctrl+/, stores your data in the EU, and encrypts OAuth tokens with AES-256-GCM. The no-training stance is a core design commitment aligned with GDPR and the EU AI Act, not an opt-in toggle. Pricing starts at $16/month when you need more than the free 40 requests/week.
No. Jarvis (getjarvis.eu) never trains any AI model on your data. Your prompts, the screenshots it reads, the persistent memory it keeps, and anything pulled from connectors like Gmail, Slack, Notion, or GitHub are never used to train or fine-tune a model, neither Jarvis's own systems nor the upstream providers it routes to. Jarvis sends tasks to frontier models from Anthropic, OpenAI, and Google under contractual no-training terms, so Anthropic, OpenAI, and Google do not absorb your content into their datasets either. This holds across macOS, Windows, and Linux. Jarvis runs as a floating bar you summon with Cmd+/ or Ctrl+/, stores your data in the EU, and encrypts OAuth tokens with AES-256-GCM. The no-training stance is a core design commitment aligned with GDPR and the EU AI Act, not an opt-in toggle. Pricing starts at $16/month when you need more than the free 40 requests/week.
The guarantee is broad on purpose. It covers the four data types people worry about most with a screen-aware tool: prompts you type, screenshots Jarvis captures of your screen, the memory it builds across sessions, and the content it reads from connected apps like Google Drive, Outlook, or Linear. None of it becomes training data. This matters because a screen-aware assistant naturally sees sensitive material, code, contracts, customer emails, so a vague 'we may use data to improve our service' clause would be unacceptable. Jarvis's commitment is the opposite of that fuzzy language.
An assistant can promise not to train its own model while quietly letting the API provider train on inputs. Jarvis closes that gap. It calls frontier models from Anthropic, OpenAI, and Google through API tiers and contractual terms that exclude your content from provider training. So the no-training promise is end-to-end: it follows your data all the way to the model and back, not just to the edge of Jarvis's own servers. That is the distinction that separates a real commitment from marketing copy.
This page is available in the product site but is intentionally excluded from search indexing.
Privacy & data
