Skip to main content
Selected work02 / 04

Shipped iOS product

Travel Agent AI

Booking capture, structured itineraries, collaboration, and practical trip operations.

Actual native views · sample dataNative iOS
01 /

Public iOS release

02 /

Version 1.38

03 /

Released August 30, 2026

04 /

Editable booking import

A CLOSER LOOK AT BOOKING IMPORT

The important step
is the review.

Extracted information should be easy to question and correct. These are the app’s native review and editing views, rendered with sample booking data.

Review before saving

Each proposed booking is selectable before it enters a trip. The sample flight exposes its title, route, cost, and confidence indicator together, so the traveler can review the proposed entry.

Native-view captures · sample data.
An isolated preview is used; saving and network access are disabled. These images are not a live AI demo.

Travel Agent AI native review screen with a selected sample AA 123 flight from JFK to LAX
Native SwiftUI / sample bookingView full size

What I owned

Product design, iOS engineering, data modeling, AI workflow design, and release operations.

Problem

Make travel planning useful after the booking confirmation, not just during destination search.

Built

Built an iOS trip assistant for booking capture and review, packing lists, calendar sync, itinerary sharing, and trip cost tracking.

Outcome

A production consumer app that turns scattered travel details into one practical mobile planning surface.

The conditions that
shaped the system.

01

Booking details arrive as screenshots, pasted text, email fragments, and manual notes.

02

Displayed itinerary times must remain stable for every collaborator rather than shifting by viewer timezone.

03

AI extraction must remain editable and understandable when source details are incomplete.

Tradeoffs made explicit

01

Structured after extraction

AI proposes typed booking fields, but the durable product model remains inspectable and editable by the traveler.

02

Wall-clock itinerary time

Stored local date and time are the shared schedule truth, preventing collaborators from seeing different itinerary times.

03

Utility over chat

The primary experience is a daily operational timeline, not an open-ended assistant transcript.

Architecture

  • SwiftUI product shell
  • Structured trip and booking model
  • AI document extraction
  • Calendar, weather, and sharing flows

Validation

  • App Store version 1.38 checked on September 4, 2026
  • Native import review supports selection and editing before saving
  • Regression tests cover booking inputs and wall-clock itinerary behavior
  • Test coverage describes implementation intent; no extraction accuracy or time-saving benchmark is claimed
Built with
  • iOS
  • SwiftUI
  • AI
  • Cloud sync
  • StoreKit
  • Calendar APIs

Shows end-to-end mobile product execution across AI-assisted extraction, App Store delivery, subscriptions, cloud sync, and everyday utility.

View on the App Store

EVIDENCE & BOUNDARIES

What this work demonstrates.

Public release + implementation review
Shipped iOS product

Apple’s public listing confirmed version 1.38, released August 30, 2026. The native booking review supports selecting and editing extracted entries before saving them into a trip.

Save AI output immediately, or ask the traveler to review it?

Decision
Present structured, editable bookings with individual selection before committing them to the itinerary.
Tradeoff
Review adds a step. In exchange, uncertain titles, locations, and local times can be corrected before they become part of the shared plan.

Scope & limitations

This demonstrates a shipped workflow, not a measured extraction-accuracy or time-saving result. Live flight status is not currently enabled. Native-view captures use sample booking data; the confidence indicator is a fixture value, not an accuracy benchmark.