Travel apps have been promising to “personalise your experience” for years. Mostly that meant showing you hotels in the city you searched for—hardly revolutionary. But over the past eighteen months, AI capabilities in travel apps have taken a genuine step forward. Some features are legitimately useful. Others are still more marketing than substance.

I’ve been testing AI-powered travel tools on recent trips around Australia and overseas, and here’s an honest assessment of what’s working.

Itinerary Planning

This is where AI is making the most noticeable difference. Apps like Wanderlog and Google’s Trip planning features now generate full multi-day itineraries based on your preferences, travel style, and time constraints.

Tell the app you’ve got five days in Tasmania, you prefer hiking over wine tours, you’re travelling with kids, and you don’t want to drive more than three hours in a day. It produces a day-by-day plan with specific stops, suggested timing, and route logistics.

The quality varies enormously. Google’s implementation benefits from its massive location database—it knows opening hours, drive times, and seasonal closures with reasonable accuracy. Smaller apps sometimes suggest places that have closed or recommend incompatible combinations (like a sunrise hike followed by a breakfast reservation 90 minutes away).

The best approach I’ve found is using AI-generated itineraries as a starting framework, then editing based on local knowledge. They’re good at logistics—optimising routes, avoiding backtracking, estimating realistic timing. They’re weaker at the subjective stuff—which beach has the best swimming versus which is more photogenic, or which restaurant is actually good versus which has the most Google reviews.

Real-Time Translation

On-device translation has improved dramatically. Google Translate’s camera feature—point your phone at foreign text and see the translation overlaid in real time—has gone from novelty to genuinely useful.

For Australian travellers in Asia especially, this is a practical improvement. Reading menus, signs, and transport information without reliable internet used to require phrasebooks or guesswork. Now it’s close to instant.

Apple’s system-level translation handles conversations well. Speak into the phone, it translates and speaks the translation aloud. I used it extensively in rural Japan last year and it handled casual conversation surprisingly well. Technical or nuanced language still trips it up.

The limitation is that translation quality depends on the language pair. English-Japanese and English-Mandarin are excellent. English to less widely-spoken languages can be patchy.

Accommodation Recommendations

This is where I’m least impressed. AI-powered accommodation matching claims to learn your preferences and suggest properties you’ll like. In practice, the recommendations feel barely better than filtered search results.

The fundamental problem is that accommodation preferences are deeply personal and contextual. I want different things from a hotel when I’m travelling for work versus a weekend getaway versus a family holiday. The AI doesn’t distinguish between these contexts well enough to be useful.

Price filtering, location filtering, and reading recent reviews still outperform any AI recommendation engine I’ve tested.

Flight and Price Prediction

AI-driven price prediction tools that tell you whether to buy now or wait have been around for a while. Google Flights shows price history and predicts whether prices are likely to increase or decrease.

These tools are moderately useful for domestic Australian flights where pricing patterns are relatively consistent. Qantas and Virgin Australia have predictable pricing curves—book early for popular routes, watch for sales on less popular routes.

International flight pricing is more chaotic and the predictions are less reliable. Too many variables—fuel surcharges, demand fluctuations, airline revenue management algorithms—make confident prediction difficult.

The most useful feature is simple price tracking with alerts. Set your desired route and dates, and get notified when prices drop below a threshold. This isn’t really AI—it’s basic monitoring—but it works.

The Strategy Behind the Tech

What’s interesting about the AI trend in travel is how it reflects broader patterns in how businesses are applying machine learning. AI strategy support for companies in the travel and tourism sector is growing because the data volumes are enormous—millions of bookings, reviews, searches, and location data points.

The companies that are using AI well in travel aren’t just slapping a chatbot on their website. They’re using it for operational improvements—dynamic pricing, demand forecasting, customer service automation—that make the experience better indirectly.

For travellers, the practical takeaway is that AI tools work best as supplements to your own planning, not replacements. Use them for logistics, data gathering, and initial ideas. Apply your own judgment for the subjective decisions that make a trip memorable.

Photo and Video Features

AI-enhanced photography in travel apps is genuinely useful. Google Photos and Apple Photos can now search your travel photos by location, content, and even specific scenes. Search “sunset” and it finds all your sunset photos from a trip. Search “food” and you get every restaurant shot.

More impressively, AI can generate highlights reels from trip photos, selecting the best shots and creating compilations. The quality is good enough for sharing, though serious photographers will still want to curate manually.

Samsung and Google phones both offer AI-powered photo editing that can remove tourists from landmark shots, adjust lighting, and enhance details. The results are remarkably good and save significant editing time.

Offline Capability Matters

One observation from testing AI travel tools across Australia: offline capability matters enormously.

Many AI features require internet connectivity, which is unreliable outside major cities. An AI itinerary planner that can’t function in areas without signal—which describes most of rural Australia—is limited.

Google Maps’ offline maps and Maps.me for offline navigation remain essential. Download everything before you leave signal range. The AI features are bonus tools for when you have connectivity, but you need offline basics as your foundation.

What’s Coming Next

The direction is clear: AI assistants that combine multiple travel functions into a single conversational interface. Instead of switching between separate apps for flights, accommodation, activities, and navigation, you’ll describe what you want to a travel AI that coordinates across all of these.

We’re not fully there yet. Current implementations are fragmented—each app does one thing reasonably well, but there’s no unified travel AI that handles everything. Expect this to improve significantly over the next year or two.

For now, the practical approach is a toolkit: Google Maps for navigation, a dedicated booking app for accommodation, an AI itinerary tool for planning, and your own brain for the important decisions. Technology is getting smarter, but it hasn’t replaced common sense and local knowledge yet.

— Lisa