Customise how your agent introduces itself and behaves.
Configured via rio configure or .env file.
rio configure
.env
One per line. The agent will ONLY answer about these.
One per line
Define what happens at each tier
Add common questions and answers the agent should know.
The agent will REFUSE to answer about these topics.
One per line — e.g., legal advice, competitor comparisons
Shown when someone asks an out-of-scope question
One per line — can be anything
One per line — optional
The tutor will focus extra help on these
Topics the tutor should NOT cover (one per line)
Configure voice translation for live shop usage. This controls both dashboard toggles and voice-mode translation behavior.
Use default unless you run multiple user profiles.
Primary language is what seller/staff should hear or speak. Customer language defaults to English and can be changed anytime.
Hotkey: press F7 in local runtime to toggle live translator instantly.
Configure multiple agents for different task types. The orchestrator will automatically route tasks to the best-matching agent based on task content. Each agent can use a different Gemini model and tool set.
Executes computer tasks: opening apps, clicking, typing, file operations.
Code analysis, debugging, file reading/writing, refactoring.
Research, analysis, explanations, complex reasoning.
Image and video generation using Imagen 3 and Veo 2.