The Rise of the Post-Search Era: How AI is Transforming Digital Interaction

In recent years, we have been evolving towards what I call a “post-search” era. Take Netflix as a paradigmatic example. With a catalog of thousands of titles and a highly diverse user base, it is almost mandatory for them to offer not only hyper-personalized title suggestions but also to propose them in an equally hyper-personalized manner. A particularly effective strategy the video streaming company uses is the customization of artwork used to represent the content. Personalizing not only the titles based on tastes but also how those titles are presented allows Netflix to maximize the click-through rate and, consequently, the time spent on the platform. It's unsurprising that, according to McKinsey, as much as 75% of the content viewed on Netflix today comes from algorithmic recommendations rather than active and conscious searches (source: How retailers can keep up with consumers)

No longer "search," in short, but "post-search." This trend is evident, though to a lesser extent, in e-commerce. Today, the most successful companies are not those with the most content but those capable of offering that content in a more finely calibrated way, focused not so much on position but prediction. The validity of this thesis can be empirically proven by the growing dominance of platforms like TikTok in the field, not only of social media but also of search itself. The AI heart of TikTok, Monolith, tailors content increasingly suited to user preferences. Once we overcome a first wave of generic content, as we interact with Monolith, the algorithm begins to reward us with increasingly relevant results, to the point of making Facebook seem more like a mass media platform than a social media platform in comparison. It is no coincidence that in recent years, there has been a change in user preferences, especially among young people, who increasingly tend to favor platforms such as TikTok and Reddit for their searches, to the detriment of Google.

“Post-search.” "Search³." Or, perhaps, even better, "anti-search." "Anti" because search implies the need (or at least the willingness) to go online and look for something. But if the results come to us without first invoking them, then the term becomes misleading. With the advent of technologies such as MUM and SGE, which bring the power of conversational natural language queries into the search experience, it's worth asking ourselves: Does it still make sense to talk about topics such as OTA ranking or search engine optimization? The web has finally begun to understand English, too, and not just coding. When Google used to say: "write for humans, don't write for algorithms," it was probably warning us. And, it was right. The bad news is that now that we've spent thirty years becoming fluent in “algorithmic lingo,” this language has become as helpful as Esperanto (meaning: not helpful at all)..

This paradigm shift suggests a radical transition in how web interfaces are conceived and used. Hypothetically, tools like ChatGPT or Claude could one day become the primary or even the sole interface for web access. I envision a future where LLMs transcend their roles as simple external data processors. Instead, these models could evolve into primary aggregators and distributors of information, particularly in industries such as travel. Such a development would signify a fundamental reversal in the traditional digital ecosystem and a move towards an unprecedented dynamic in terms of interface. Today, OpenAI's APIs power Expedia and Booking UIs, but tomorrow, it could be OpenAI’s UI powered by OTA’s APIs.

This evolution would radically transform the architecture and interface of the World Wide Web as we know it. Browsing through multiple sites or proprietary applications may become obsolete. In the (near?) future, users could prefer to interact exclusively through a single, all-knowing conversational platform. A transition from an informative web, based on HTML user interfaces and mobile apps, to a generative web, supported by sophisticated LLM algorithms and driven by a new AI (post) search paradigm. A centralized AI assistant powered by a myriad of APIs, with the World Wide Web as the underlying infrastructure and large language models as the user interface.

This shift could significantly transform the traditional reliance on websites and applications to a more unified and composite web interaction. Ideally, a user could scan the label of a bottle of wine, receive an immediate assessment of its qualities, and then seamlessly move on to queries about the provenance of the wine, suggestions for travel destinations related to its origin, recommendations for activities, and even the best way to communicate with locals, all within a single interface. This evolution would be not just a technological advancement but also a behavioral redefinition of how we interact with the digital world, converging specialized functionality into a cohesive, AI-centric user experience. It'll mark the end of what I call the "logistics" of travel planning (Hallelujah!).

If today we open a thousand browsers and apps, tomorrow we will open a single conversational platform. The practice of sifting through a multitude of websites and applications is, in hindsight, only a convention born out of necessity and, therefore, not immutable, we know that. The web began with sites, which were then indexed by search engines. This led to meta-search engines, which indexed the search engines. The advent of mobile technology led to the creation of apps designed to improve this frustrating user experience, forcing us to develop app stores to organize this chaotic ecosystem. We have built platforms to aggregate other platforms, which aggregate other platforms. With over a billion websites and nine million apps, I believe that the digital world is ready for a change, as transformative as it is necessary, towards a web in which decision-making processes and digital interaction are simplified and, upon closer inspection, more aligned with human cognitive and conversational models.

It's just funny that it took AI to make search more human.

Isn’t it?