---
title: "Building Multi-City Itineraries: When Simple Search Becomes Complex Planning"
description: "Teaching the agent multi-city trips — combinatorial complexity of multi-leg itineraries, conversational UX for complex options, and multi-segment booking flows."
canonical: https://nowah.xyz/blog/launching-multi-city-itineraries-complex-planning
lastModified: "2026-08-07T08:24:30.291Z"
---

# Building Multi-City Itineraries: When Simple Search Becomes Complex Planning

Teaching the agent multi-city trips — combinatorial complexity of multi-leg itineraries, conversational UX for complex options, and multi-segment booking flows.

A traveler said: "Paris for 3 days, then Rome for 4, then Barcelona for 2." She expected a single answer. What she actually asked for was a combinatorial problem that makes a simple point-to-point search look trivial.

Three cities mean three flight segments. If each segment has ten viable options, the total combination space is one thousand possible itineraries. The agent cannot show a thousand options. It needs to reason about the entire trip as a connected whole, optimize across segments rather than within them, and present a handful of curated options that represent genuinely different tradeoffs.

## The combinatorial explosion

![Illustration for this section](https://pics.nowah.xyz/website-media/launch-022-img-1.webp)

Point-to-point flight search is linear. More options means a longer list, but each option is evaluated independently. [Multi-city](/blog/multi-city-flight-booking-ai-agents) search is multiplicative. Each segment's options combine with every other segment's options, creating an explosion of possibilities.

The real complexity goes beyond multiplication. Segments are not independent. The Paris-to-Rome flight must depart after the traveler finishes their Paris stay. The Rome-to-Barcelona flight must depart after their Rome stay. [Connection quality](/blog/airport-intelligence-connection-quality) between segments matters: a tight transfer between a late Paris departure and an early Rome hotel check-in creates a stressful experience even if each leg is individually optimal.

Our agent handles this by decomposing the request into segments, searching each segment independently to build the option space, then evaluating combinations holistically. The evaluation considers total price across all segments, total travel time including transfers, connection quality between each city, and carrier consistency for travelers who prefer booking all legs on the same airline family.

## Presenting options conversationally

The challenge is not just finding good multi-city itineraries. It is presenting them in a conversation without overwhelming the traveler. A traditional OTA might show a grid of options across segments. In a conversational interface, that grid becomes a wall of text.

We present multi-city options as curated packages: "Here are three itinerary options for your Paris-Rome-Barcelona trip. Option one optimizes for price. Option two optimizes for convenience with the shortest total travel time. Option three uses a single airline alliance for all legs." Each option is a complete itinerary, not a mix-and-match grid.

If the traveler wants to modify one segment while keeping the others, the agent handles that as a conversation: "I like the Paris-Rome flight in Option one but prefer the Rome-Barcelona timing in Option two. Can you combine them?" The agent understands this as a partial recomposition and searches for the optimal third segment to complete the modified itinerary.

## Multi-segment booking

![Supporting diagram](https://pics.nowah.xyz/website-media/launch-022-img-2.webp)

Booking a multi-city itinerary is more complex than booking a single flight. The traveler needs to confirm all segments together because changing one segment might invalidate the timing of another. The booking review modal shows all segments in a stacked view with total price, total duration, and connection quality scores.

If a price changes on one segment between search and booking, the agent recalculates the total and alerts the traveler before proceeding. The multi-layer [idempotency](/blog/idempotency-travel-booking) checks handles multi-segment bookings as a coordinated transaction: either all segments book successfully, or the entire booking is rolled back.

Multi-city itineraries have higher average order value than single-segment bookings but lower completion rates. The [decision fatigue](/blog/decision-fatigue-travel-science) from managing multiple segments causes some travelers to abandon the process. Our conversational approach, presenting curated options rather than a grid of combinations, directly addresses this fatigue. The agent absorbs the complexity so the traveler does not have to.

---

Nowah is an AI travel agent that searches and books real flights and hotels through conversation — no filters, no thirty open tabs. [Plan your next trip](https://app.nowah.xyz).