LlamaIndex destination brief

Compare remote-work destinations with travel, web, and maps evidence.

What you’ll build

A LlamaIndex FunctionAgent that discovers reachable cities for a short remote-work trip, researches practical context, checks local coworking access, and recommends one destination with explicit tradeoffs.

Default brief: Starting from JFK, find three promising destinations for a four-day remote-work trip and recommend one with evidence and source URLs.

How the agent works

  1. Explore reachable cities. Travel explore search supplies indicative flight options from the origin airport.
  2. Test practical fit. Web search checks remote-work facts and maps search verifies nearby coworking access.
  3. Compare explicitly. The FunctionAgent gathers evidence from every relevant tool before choosing one destination.

Core agent setup

This excerpt shows the LlamaIndex FunctionAgent and its required first tool call. The complete script also configures the model, validates credentials, sets timeouts, and saves the destination brief.

agent = FunctionAgent(
    tools=[
        travel_explore_search(provider="llamaindex", api_key=api_key),
        web_search(
            provider="llamaindex",
            allowed_engines=["google_light", "bing"],
            api_key=api_key,
        ),
        maps_search(provider="llamaindex", api_key=api_key),
    ],
    llm=llm,
    initial_tool_choice="required",
    streaming=False,
    system_prompt=(
        "Gather evidence from each relevant tool, make the comparison explicit, "
        "and include source URLs."
    ),
)
result = await agent.run(PROMPT)

View the complete main.py →

Run the recipe

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

uv run --isolated --no-project --with 'serpapi-search-tools[llamaindex]' --with python-dotenv --with llama-index-llms-openai \
  cookbook/llamaindex/main.py

Optional controls: OPENAI_MODEL, COOKBOOK_PROMPT, and COOKBOOK_OUTPUT_DIR.

Tool use starts immediately: The agent sets initial_tool_choice="required", preventing a plausible-sounding destination answer before live travel evidence is collected.

Inspect the result

The recipe writes cookbook-output/llamaindex-destination-brief.md. Expect three candidate destinations, indicative travel evidence, practical remote-work context, coworking access, tradeoffs, one recommendation, and source URLs.

Replace the origin, trip length, or decision criteria through COOKBOOK_PROMPT.

SerpApi tools used

travel_explore_search identifies reachable destination candidates, web_search checks practical remote-work facts, and maps_search verifies local coworking access. Together they help the agent compare reachability, practical fit, and nearby facilities as separate decision criteria.

Framework pattern: The FunctionAgent workflow follows LlamaIndex’s basic agent example and uses three complementary SerpApi tools.