Conversational multi-agent travel planner. Describe a trip in plain language and get live flight status, hotel research, a day-by-day itinerary, and a budget-aware plan.
https://github.com/nilaymallikk/RoamAI/raw/main/docs/ROAMAI_DEMO_1.mp4
- Chat UI — FastAPI + Jinja frontend with browser-local journey history (24h)
- Multi-agent pipeline — sequential LangGraph agents for flights → hotels → itinerary → final answer
- Live flights — AviationStack route search with city/country → IATA resolution
- Hotel research — Tavily web search for stays
- Memory — PostgreSQL checkpointer for thread-based conversation state
- LLM — Groq (
llama-3.3-70b-versatile) for itinerary and final response
Flow: user message → flight_agent (AviationStack) → hotel_agent (Tavily) → itinerary_agent (LLM) → final_agent (LLM) → structured travel plan.
| Layer | Stack |
|---|---|
| API / UI | FastAPI, Jinja2, vanilla JS/CSS |
| Agents | LangGraph + LangChain |
| LLM | Groq |
| Tools | AviationStack, Tavily |
| State | PostgreSQL (langgraph-checkpoint-postgres) |
1. Install
pip install -r requirements.txt
# or: uv sync2. Environment — create a .env in the project root:
GROQ_API_KEY=...
DATABASE_URL=postgresql://user:pass@host:5432/dbname
AVIATIONSTACK_API_KEY=...
TAVILY_API_KEY=...
DEFAULT_ORIGIN_IATA=DEL # optional3. Run
python app.py
# or: uvicorn app:app --host 0.0.0.0 --port 8000 --reloadOpen http://127.0.0.1:8000. Health check: GET /health.
Docker
docker build -t roamai .
docker run -p 8000:8000 --env-file .env roamaiPOST /api/travel
{
"message": "Plan a 7-day Japan trip from Delhi under a mid-range budget",
"thread_id": null
}Returns answer, flight_results, hotel_results, itinerary, thread_id, and llm_calls.
app.py # FastAPI routes + static UI
backend.py # LangGraph travel agents
tools/
flight_tool.py # AviationStack + IATA resolution
tavily_tool.py # Hotel / web search
templates/ # Chat UI
static/ # CSS, JS, images
docs/ # Demo video + architecture diagram
MIT — see LICENSE.