FAQ

Privacy

When is the agent active? The agent is active only when you summon it via the shortcut key (e.g., tapping the option key), highlight a piece of text, or activate the context selection tool.

Does it have access to my personal files? It does not have permission to create, view, modify, or delete any of your personal files.

Does it have access to information on my apps? The agent does not record or store any application-specific information when it takes action in any application on your behalf. For example, when you ask it to share your current application window to an ongoing video conference meeting, the agent does not use, read, or store any information in the application or the meeting.

Simular is wants to use google.com to Sign In. If I give consent, to what information will it have access? This sign in method uses Google Sign In and your Google account name (e.g., @gmail.com) for authentication.

Simular is requesting permission to access my Google Calendar. If I give consent, to what information will it have access? This permission is required for the agent to use the Google Calendar API to check and set events on your calendar. If you give consent, the agent will be able to 1) read events on your existing Google Calendars, 2) create a new secondary Google Calendar named “simular.ai” and create, edit, and delete events on that secondary calendar. The agent will not create, edit or delete events on your existing calendars.

What data do you collect? The following usage data is stored only on your device (not transmitted through the internet):Query and contextThe following usage data is stored on our secure servers:Name of the top-level application from which the agent was summoned.Length of prompt and context.Non-PPI (personal identifiable information) words in the query.Action that was chosen by the agent.

Safety

How safe is mouse control mode? Mouse control mode is a new technology that lets the agent complete tasks autonomously based on your request and permission. In mouse control mode, the agent is able to move the mouse and click objects on the screen, just as you would. It will only click on objects that are necessary to complete the task. By default, mouse control mode is disabled until you enable it in the Preference menu. To improve safety, the default setting in mouse control mode will require your approval for each mouse action in an action sequence, which you can give by pressing the [right arrow] key. You can turn off this additional safety check to let an action sequence flow autonomously.

Will the agent ask for permission before taking action? Yes, the agent will ask for your confirmation before executing any sequence of actions, except for actions that only retrieve information (such as responding via language, or checking your Google Calendar). The default setting in mouse control mode will require your approval for each mouse action in an action sequence, which you can give by pressing the [right arrow] key.

Can I stop the agent’s current action? If the action sequence consists of more than one step, then you can cancel the next actions by moving the mouse to one of the four corners of the screen.