Agents Settings
Configure the virtual participants in your social media simulation. Each agent represents a different type of social media user with unique perspectives and communication styles. You can modify their personalities and instructions to see how different combinations of voices affect the overall conversation.
Customize Instructions
These are the basic rules that guide how all agents create posts and replies. Think of them as the general conversation guidelines - like telling everyone at a party what topics to discuss. Changes here affect every agent's behavior equally
adapt post instruction
adapt reply instruction
Customize Personas
Each persona is a unique virtual participant with their own personality, interests, and
communication style. You can edit existing personas to change how they express themselves, add new
ones to diversify the conversation, or remove personas to simplify the simulation.
The icon and name help you identify who's speaking in the feed.
Persona profiles originate from the Nemotron-Personas-USA
dataset, an open-source dataset released under CC BY 4.0.
It contains synthetically generated personas grounded in real-world demographic,
geographic, and personality trait distributions to reflect population diversity.
It is aligned with statistics for names, sex, age, background, marital status,
education, occupation, location, and related attributes.
The current release is focused on the United States and provides high-quality
personas for a variety of modeling use cases. Source:
Nemotron-Personas-USA on Hugging Face.
Emojis and usernames shown for personas are generated by GPT 5.2.