groups Multi-User & Multi-Agent

Deepthink can be orchestrated for multiple users and multiple agents by isolating credentials and storage per tenant. Assign separate MongoDB databases and Neo4j databases per tenant to prevent data leakage.

In multi-agent setups, each agent can run with its own LLM provider and configuration while sharing a controlled knowledge layer.

For user scoping, create a UserAgent instance and set user_authority. If the user is an admin, use 1.0. If the user is regular and their data should remain private, set it to 0.

from analogai.deepthink.application.value_objects.user_agent import UserAgent

admin_agent = UserAgent(
    user_id="admin-user",
    agent_id="primary-agent",
    user_authority=1.0,
)

private_user_agent = UserAgent(
    user_id="user-123",
    agent_id="primary-agent",
    user_authority=0,
)