Travel search giant KAYAK launched Ask AI in late April, a conversational planning tool.
While this may sound fairly standard in today’s AI-powered world, what sets KAYAK’s product apart is that it keeps a live, traditional search results page on screen while the chat is happening.
This means there are no redirects and no starting over. Travelers can describe what they want in plain language, refine the brief mid-conversation, and watch bookable options update in real time alongside the chat.
According to Matthias Keller, Chief Product Officer at KAYAK, the insight behind it came from a clear signal in the data:
“In more than 100,000 AI conversations a month with our AI chat products, we saw that AI is especially helpful at the exploratory start of planning, but travelers are not yet ready to make a decision from just a handful of AI suggestions alone.”
“They still want to see the full set of live results, compare options, and use the filters and controls they already know and trust.”
The gap between inspiration and booking, KAYAK found, was more of a user experience problem than a content one.
The Behavior Behind the Build
KAYAK introduced natural language search back in October 2025. The data from that rollout showed something telling: travelers were using the conversational entry point to start shaping a trip, then reverting to traditional search when it came time to compare and commit.
Keller frames this as less of a flaw and more a reflection of how planning actually works:
“AI fits very naturally at the beginning of the journey, when people are still shaping the trip and often do not yet have fixed dates, a firm destination or a fully formed brief.
“But when they get closer to a decision, travelers still want to compare prices, timings and policies in a structured way before they book.”
His conclusion lands as a pointed observation for the broader AI travel space:
“The right model is not chat instead of search, but chat working alongside search.”
Indeed, a significant portion of AI travel tools have positioned themselves as replacements for the traditional search-and-filter experience. KAYAK’s argument, backed by its own usage data, is that the approach misreads where travelers actually need help.
Trust and the Design Decision
Any conversational interface in travel has a trust problem to solve. Booking a flight or hotel involves real money and real consequences, and a chatbot returning a handful of suggestions with no broader context to cross-reference can feel like a leap of faith.
KAYAK’s answer was to ground Ask AI in live, bookable results rather than curated recommendations.
As a user refines their trip in the chat, the results page on the same screen updates in real time.
According to Keller, this “keeps the experience conversational while still giving them the transparency to compare options properly.”
The launch timing, right as summer searches ramp up, reflects a deliberate read on how traveler behavior has shifted. People are taking longer to commit, tracking prices more closely, and scrutinizing cancellation policies earlier in the process, as Keller explains:
“Ask AI is built for that more iterative behavior because travelers can describe exactly what they need, adjust the brief as they go and see live options update instantly without starting over each time.”
A Feature or a Shift?
The bigger question is whether Ask AI is a product addition or the beginning of something more structural. In Keller’s opinion, it’s both:
“It is both a feature travelers can use today and a signal of where travel search is heading longer term.
“Over time you can expect the experience to become more conversational and more personalized, but we see that as an evolution of metasearch rather than a replacement for the core tools travelers trust when it is time to book.”
That’s a careful position, and probably the right one. The travel industry has been through enough AI hype cycles to know that replacing what works is a harder sell than improving how people get to it.
KAYAK doesn’t appear to be trying to reinvent the booking experience; instead, it is trying to make the road to it considerably less bumpy.
