Jet4Holiday
Inspiration
Travel planning is fragmented: one site for flights, another for hotels, countless tabs for restaurants and activities. We wanted to eliminate the juggling. Jet4Holiday lets specialized AI agents handle each part of the journey independently—then collaborate to produce a single, coherent, personalized itinerary.
What it does
Jet4Holiday is an agent-to-agent (A2A) travel planner. Instead of one generic assistant, it deploys multiple domain-focused AI agents (flights, hotels, food, experiences, budgeting) that communicate and negotiate to deliver:
- Optimized outbound + return flight and lodging pairings
- Curated, context-aware food and activity suggestions
- Adaptive day-by-day itineraries that rebalance when constraints shift
- Real-time budget tracking, savings insights, and trade‑off awareness
How we built it
- Agent Framework → Each travel domain runs as an autonomous, async agent with its own decision logic.
- Collaboration Layer → Agents exchange structured proposals to resolve timing, budget, and dependency conflicts (e.g., late arrival vs. early tour).
- Data Sources → External APIs for flights, hotels, maps/places, and local experiences with fallbacks for resilience.
- Frontend → A clean interface for preferences, itinerary viewing, and interactive refinement.
- Backend → Orchestrates agent lifecycles, mediates negotiations, aggregates results, and personalizes output.
Challenges we ran into
- Tuning the trade-off between speed and negotiation depth.
- Reconciling conflicting user priorities (cheapest vs. shortest vs. premium options).
- Architecting a scalable orchestration layer without creating a central bottleneck.
- Designing UI that feels simple while showcasing multi-agent intelligence.
Accomplishments that we're proud of
- Achieved smooth agent-to-agent collaboration producing cohesive itineraries.
- Built a flexible framework that can onboard new agents (e.g., insurance, local transport) rapidly.
- Established a modern, extensible brand identity—Jet4Holiday feels global and product-ready.
- Delivered adaptive itineraries that react to timing, budget shifts, and resource availability.
What we learned
- Multi-agent systems unlock richer planning—but orchestration logic is the hardest part.
- Trust and transparency (showing “why this was chosen”) matter as much as raw optimization.
- Travel planning blends hard constraints (time, budget) with emotional intent (vibe, uniqueness).
What's next for Jet4Holiday
- Add new agents: local transport, insurance, concierge, carbon impact.
- Integrate real-time signals (weather, delays, local events, dynamic pricing).
- Launch a mobile-first companion for in-trip adaptation.
- Enable collaborative trip planning with shared agent sessions.
- Move toward fully auto-generated, continuously adaptive itineraries.
Built With
- amadeus
- auth0
- css
- gemini
- google-cloud-a2a
- google-places
- html
- pydantic
- python
- sql
- tripadvisor
- uvicorn
- voice-activation

Log in or sign up for Devpost to join the conversation.