Order the agents ran in:
  findFlight read [traveler, destination] and wrote AI-101 Mumbai to Lisbon
  findHotel read [traveler, destination] and wrote Hotel Alfama
  findActivity read [traveler, destination] and wrote Tram 28
  write read [hotels, activities, destination, flights] and wrote Fly AI-101, stay at Hotel Alfama, ride Tram 28.

Shared state at the end:
  traveler = Ankur
  hotels = Hotel Alfama
  activities = Tram 28
  destination = Lisbon
  flights = AI-101 Mumbai to Lisbon
  plan = Fly AI-101, stay at Hotel Alfama, ride Tram 28.

The prompt the planner was sent (last model call):
  Write a one-line itinerary for Lisbon. Flight: AI-101 Mumbai to Lisbon. Hotel: Hotel Alfama. Activity: Tram 28.

Model calls: 4