I guess there are bots here making all these political comments to diverge the discussion and not talk about what really matters: is the model good to do actual work?
Looks interesting, but I feel that as models get better and better, the need to have 'personas' for each agent is actually a drawback compared to having a well-defined starting plan and objectives, because more 'intelligent' models will just figure out a more efficient way to do what they need to do.