[julien-nc]

• agent user management
• allow @mentions to agents
• responses in 1-1 rooms
• reactions
• agency confirmations

Questions

• There's a conflict with the "single source of truth" aspect:
  • We want the human user to see the conversations in the Assistant UI as well (so we kind of have to schedule the chat tasks as the human user)
  • We want the actions to be performed as the agent

🤖 AI (if applicable)

• The content of this PR was partly or fully generated using AI (N/A)