AI Search needs to evolve quickly, as it is currently a loss-making black hole: neither ChatGPT nor Perplexity is profitable, and Gemini is dramatically impacting Google’s PPC revenues, with a negative knock-on effect on profits.

The entire ethos of AI search is to guide customers to an answer more quickly, generating high-intent customers who are better suited to a “Cost per Acquisition” (CPA) or retail model like Amazon.

Anthropic signalled the first major move in this direction with the creation of its MCP (Model Context Protocol), which is simply a standardisation “Brick” that suppliers must connect their API’s too, so that “Large Language Models” (LLMs) can hoover them up to build super stores with.

However, last month, Google went a step further, launching its new UCP (Universal Commerce Protocol), a standardised way for retailers to connect to Gemini or other LLMs to enable agentic commerce, allowing AI agents to perform product discovery, manage carts, checkout, payments, order tracking, and post-purchase flows.

The key point is that UCP shifts the transaction point from the retailer’s website to within the AI search itself to optimise user intent and reduce booking friction caused by the current need to visit the retailer’s site. The retailer remains the merchant of record and receives the customer’s details; however, fewer human customers will visit the retailer’s site, as most traffic comes from APIs and machines.

So, how will this impact travel businesses?

Clearly, businesses must set up a “Merchant Centre” account with Google and deploy their own UCP servers, though this process is relatively simple. However, what businesses often need from consultants like my own “Neural River” is insight into the wider implications of this change and guidance on what they should prepare for next.

For example, AI Search engines are “conversational” and, unlike travel sites, don’t use “order form” searches to obtain the key search data such as departure date, airport, destination, etc. Instead, they will handle a word-based search, such as “Find me cheap all-inclusive deals to Alcudia in Majorca, for a family of 2 adults and two children, departing on the 12th of July from Manchester airport”.

Now, you may spot a usability issue straight away. That’s a lot of typing!

This is why all AI search engines are rapidly shifting to promote voice search, where users simply chat with the search engine about their needs, as demonstrated by last week’s Uber Eats new voice service.

Initially, there will be some resistance to this move among the “Betamax” generation of 20-30-year-olds, who have been trained only to interact via typing, but the convenience factor will quickly change user behaviour.

The shift to voice will also broaden the query string and introduce more holiday requirements that responses need to meet, such as reliable Wi-Fi, kids’ clubs, and quality restaurants and bars near the hotel.

However, the biggest change is that AI engines will demand “Recommendations” rather than thousands of options, as they want to guide customers to a decision and complete the transaction quickly. So less is more

Also, as “Digital Twins” develop on customers’ phones that know all their key preferences, tagging properties as suitable for XYZ and creating “lookalike” customer groups, where a “Customer like you enjoyed this holiday” will become a key conversion driver.

Therefore, exploring new methods to gather rich user content like Travel Voices’ “Digital Holiday Books” is important for both driving “Friend get Friend” referral sales and creating an extensive database of customer reviews, where new users can talk via AI to previous customers about their holidays.

Travel search is poised for a major evolution, with voice search and recommendations replacing the standard online search forms that are the core of every current travel website.

Are you ready for this shift?