PagesThe ideaPossible forms of the systemTechnical base optionsOpen design questionsLimits and securityPersonalities and profilesUse scenariosScenario 1: Local use: me and my agentsScenario 2: Me, my agents and guestsScenario 3: Company: one departmentScenario 4: The whole companyScenario 5: GamesScenario 6: Customers and suppliersPossibilities for companiesPossibilities for gamesThe INEMA case (for discussion)Sources

Personalities and profiles

In the multiplayer space, each member has a profile. For people, the profile says what they do. For agents, it also says how they think and speak. The source video shows both kinds of profile, and that is what makes it possible to call the right agent for each task.

1. Person profile

Field Example
Role video editor, operations manager, presenter
Responsibilities what is theirs in the workflow
Current goals goals for the week or month
Work context tools they use, hours, how they prefer to be addressed

The video shows that each person fills this in on joining, in a kind of interview with the agent, and the agents then use this information (for example, to know whom to chase for a task).

2. Agent profile

Field What it defines
Name and owner who it is and whom it belongs to
Specialty where it is good (text, code, design, finance)
Model and reasoning level how much it "thinks" before answering, and how much it costs
Where it runs on the owner's machine (subscription) or in the cloud (per use)
Personality temperament: direct, careful, provocative, didactic
How it thinks method: checklist, questions before acting, always with sources
Voice tone and vocabulary, formal or informal, short sentences
Quota how much of the subscription remains (the video shows "97% remaining")
Status online or offline

3. Intelligence levels

The video shows hosted agents with three levels. The cost difference is the one it reports:

Level Typical use Relative cost (according to the video)
Fast administrative tasks, triage, short answers 1/10 of the top level
Pro reasoning, analysis, decisions half of the top
Top complex systems, engineering, design reference (the author says he uses the strongest model available)

Possibility: build the team by level, with a fast reception agent, pro agents for analysis and a top one only for heavy work. This controls cost without losing quality where it matters.

4. Personality archetypes

Personalities that make sense in a shared space:

Archetype Role in the channel Where it appears
Leader / orchestrator Receives the request, distributes it to the others and joins the result All scenarios; in the video, the workspace's lead agent
Specialist Does one thing, done well (design, code, text, data) Department and company
Project manager Chases deadlines, does the daily check-in, flags who is late Department and company
Council Optimist, devil's advocate and neutral discuss before the human decides Important decisions
Reviewer Checks the other agents' work before it reaches the human Any flow with publishing
Receptionist First contact with a customer or supplier, in the brand's tone Scenario 6
Game master Narrates, applies rules and keeps the world coherent Scenario 5
NPC Character with its own story, motivation and memory Scenario 5; each player talks to it in private [14]
Tutor Teaches, chases delivery and unlocks the next stage Course as a game
Owner's twin Speaks like the owner, with their knowledge, when the owner is away Scenario 2; the video describes the wish that colleagues could "write to my AI system"

5. How personality combines with the rules

6. Risks

7. Open

  1. Who defines the personality: the agent's owner or the channel administrator?
  2. A single personality or one per channel?
  3. Is it worth having a "catalog" of ready-made personalities (leader, reviewer, council) for people getting started?
  4. Can the owner's twin answer alone or does it always warn the owner first?