CrewAI collaborative trip planner

Research and edit a trip plan with typed travel search.

What you’ll build

A two-agent CrewAI trip planner. A researcher gathers current flight, hotel, and neighborhood evidence; a decision editor turns that packet into a practical four-night plan with an explicit budget.

Default brief: Plan Kyoto for two adults, flying SFO to KIX roughly 120 days from now. Keep flights and lodging under USD 4,000 and prefer a walkable, transit-friendly neighborhood.

How the agent works

  1. Search exact travel inputs. The researcher uses airport codes, generated dates, occupancy, and currency with flights and hotels search.
  2. Ground the neighborhood. Maps search checks nearby places and gives the planner location-level evidence.
  3. Edit the decision. A second agent selects one flight and hotel, totals the estimate, and surfaces assumptions or price volatility.

Core agent setup

This excerpt highlights the typed travel tools and the two-agent handoff. The full script defines the research and planning tasks, runtime dates, model, and saved report.

travel_tools = [
    flights_search(provider="crewai", default_params={"currency": "USD"}),
    hotels_search(provider="crewai", default_params={"currency": "USD"}),
    maps_search(provider="crewai", default_params={"gl": "jp"}),
]
researcher = crewai.Agent(
    role="Travel search researcher",
    goal="Collect current flight, hotel, and neighborhood evidence.",
    llm=llm,
    tools=travel_tools,
)
planner = crewai.Agent(
    role="Travel decision editor",
    goal="Turn verified options into a practical, budget-aware trip decision.",
    llm=llm,
)
crew = crewai.Crew(
    agents=[researcher, planner],
    tasks=[research_task, planning_task],
    process=crewai.Process.sequential,
)
report = str(crew.kickoff())

View the complete main.py →

Run the recipe

Set SERPAPI_API_KEY and XAI_API_KEY in the repository-root .env, then run:

uv run --isolated --no-project --with 'serpapi-search-tools[crewai]' --with python-dotenv cookbook/crewai/main.py

Optional controls: XAI_MODEL, XAI_BASE_URL, COOKBOOK_PROMPT, and COOKBOOK_OUTPUT_DIR.

Dates stay runnable: The default prompt calculates departure and return dates at runtime. Supply COOKBOOK_PROMPT when you want a different origin, destination, party, or budget.

Inspect the result

The recipe writes cookbook-output/crewai-trip-plan.md. Expect one recommended flight, one hotel, an estimated total, a compact four-day outline, source URLs, and assumptions that could change the price.

The researcher can take up to seven iterations; the editor gets a shorter three-iteration pass over the collected evidence.

SerpApi tools used

flights_search checks routes with exact airport identifiers and dates, hotels_search uses the stay dates and occupancy, and maps_search grounds the neighborhood recommendation in nearby places. Their typed inputs help the researcher collect options that match the trip constraints before the planner makes a decision.

Framework pattern: The researcher-and-planner roles follow CrewAI’s Trip Planner example and use SerpApi flights, hotels, and maps tools.