Workspace content
Documents, notes, project files, and references that the user intentionally adds or connects for assistant use.
Privonaut Privacy
A personal AI workspace needs clear data boundaries.
Privonaut is designed around a simple idea: a personal AI assistant should work with the user’s own documents and tasks without turning that workspace into an uncontrolled data trail. Because the product is built for personal knowledge, privacy is not an add-on; it is part of the product model.
Privonaut will make data use visible in the interface: what has been added to the workspace, what the assistant can search, which tools are enabled, and what information is used to answer a request. The exact controls may evolve as testing continues, but the direction is fixed: clear choices, limited access, and no vague hidden defaults.
What Privonaut may use
Only the information needed to make the workspace useful.
Workspace settings
Preferences, enabled tools, model choices, retrieval settings, and task templates that define how the assistant should behave.
Operational metadata
Technical information needed to run the product, diagnose errors, enforce limits, and maintain reliability.
User-provided feedback
Ratings, corrections, or comments the user chooses to provide to improve workflow quality and assistant behavior.
How the workspace is protected
The product is designed around controlled access, not unlimited memory.
Explicit sources
The assistant should work from sources the user adds, selects, or enables, rather than silently scanning everything on a device.
Retrieval over retraining
Documents can be used as searchable context without needing to permanently retrain the underlying model on every file.
Tool boundaries
Actions such as file work, code execution, or integrations should run with limited permissions and clear user control.
Visible settings
Important privacy-related choices should be shown in plain language instead of hidden behind technical configuration.
Private model options
Where supported, Privonaut is intended to offer more private model and deployment options for users who need stronger control.
No magic claims
Privonaut will not claim that AI can make every workflow perfectly private or risk-free. Boundaries will be explained clearly.
Important boundaries
A personal AI workspace is powerful, so the limits must be honest.
Privonaut is not a promise that every AI action happens only on one physical device in every configuration. Some setups may use local models, some may use private server infrastructure, and some features may require optional external services or integrations. When that happens, the product should make the trade-off understandable before the user relies on it.
Users should avoid connecting documents or tools they do not want the assistant to use. The safest workspace is a deliberate workspace: selected sources, clear permissions, and regular review of what the assistant can access.