I need a working AI-powered travel planner that can read a user’s budget, preferred destinations, and travel dates, then generate a door-to-door business-trip itinerary and handle the heavy lifting—from research to booking prompts—in one seamless flow. The product should behave much like the roadmap shown here: https://www.roadmapflow.com/roadmaps/fabefcd1-3bb5-4538-a2c3-4660ef0fd1f4, but tuned specifically for corporate travel scenarios. Core flow I’m after • Conversational intake: the system asks clarifying questions, captures budget limits, dates, and destination preferences. • Real-time planning: it assembles flights, hotels, ground transport, meeting slots, and generates a polished agenda. • Smart recommendations: swaps options automatically when flights fill up or budgets shift, always keeping costs in range. • One-click actions: deep links or API calls that push confirmed choices straight to booking engines or an admin inbox. • Export & sync: PDF/HTML shareable itinerary plus calendar and expense-report feeds. Must-haves • Business-trip focus from day one (leisure and adventure can stay in the backlog). • Proven travel data sources—Amadeus, Skyscanner, Google Places, Rome2Rio, or your preferred stack. • OpenAI (or comparable) LLM workflow orchestration; feel free to wire in LangChain, Python, Node, or a React front end. • Clean, well-commented code and a short Loom or written walkthrough showing setup and demo. Acceptance criteria 1. A user supplies budget, destination, and dates via chat or form and receives a complete business itinerary in under one minute. 2. All itinerary items show live prices pulled from at least two external travel APIs. 3. “Book now” or “Request booking” links function for flights and hotels. 4. Code installs locally via README in <30 min and deploys to a cloud host with provided instructions. 5. Delivery, including source, demo account, and documentation, is handed over within one month of contract start. If you’ve built similar AI assistants, especially around travel or logistics, I’d love to see a demo or repo link when you respond.