Every year, Google holds its I/O event where it announces the latest releases and updates. This year was an anticipated one because Google has been particularly quiet about their AI offering, leading people to speculate that they may be working on something big! Well, it looks like they weren’t when it comes to Search.
I will be focusing on Search which, in my opinion, has failed to deliver anything meaningful for the industry this year, despite Google’s best attempts at making us think so. We will be talking away the “hype” and looking at what has actually changed and how it will shape our search offering going forward.
But first, a little personal note on the overall event.
This event did release some seemingly interesting news overall and while I have plenty of opinions about everything, and some things on the hardware side appear genuinely interesting, overall, it didn’t really meet my expectations.
Judging by the number of [Applause] placeholders that were never fulfilled, it’s fair to say that Google was a lot more optimistic about this one than the people attending or watching. I think Apple Insider probably captured this very well saying: “…over and over again, presenters paused for applause that just did not come.”
And assuming the audience present at the event is likely some of the most engaged with Google, that is not a great start! This was also reflected in the poor view to engagement ratio on their YouTube video being among the weakest the event has had.
Here are the changes that Google announced that might affect Search Marketing and SEO.
AI Mode and Overviews merged and upgraded.
A faster, more capable model handling AI Mode answers. They also merged AI Overviews and AI Mode into one flow, so you move from a normal query, to a results page with an AI Overview, to a follow up in AI Mode without leaving the page, and that part is live worldwide now. The practical effect is that AI Mode can do more in a single query, which encourages longer and more complex user inputs. Long-tail keyword volume keeps compressing as people do more research without clicking through to any website.
As this experience gets better and more people start adopting it, traffic from informational searches will keep declining. Blogs and articles that were once the backbone of growth in SEO will not attract as much traffic anymore. This has been happening for the past 2 years or so as Google has been triggering more AI Overviews. The value of content marketing has undoubtedly declined or at least meaningfully changed.
At Bigwave we shifted a long time ago to ensure the content our clients get captures traffic from Social, AI Referrals, and Search, supplementing some of the decline.
We use content as a strategic asset to improve visitor experience, increase referral traffic and links and position the client’s website as a source of truth for both users and the AI that relies on it.
AI has also given us additional leverage to create content that is more creative and diverse without breaking the client’s budget. The bottom line is that, as old channels constrict, new ones open!
Redesigned search box and what they forgot to mention
Optimisation for clean head terms gets less useful, content that answers fully formed questions around well-defined topics, in plain language gets more useful.
The new search box accepts text, images, files, videos and Chrome tabs as inputs. Queries get longer, more specific and increasingly multimodal in origin. GSC will keep showing the text portion of queries, but a growing share of search intent will arrive with reference material we cannot see, embedded in the Large Language Model.
Despite everyone jumping on this as the definite “SEO killer”, Google “forgot” to mention that this search box is the entry point to AI Mode, not a replacement for the standard search field everyone already uses. This is very important because it means that most users will still search as usual and get the normal search results.
So, is it all hype? No. It’s kind of like the “floating island” equivalent from Apple. It just makes things a bit better and drives more people to try AI Mode.
Generative UI and mini apps in search
AI Mode builds custom interfaces, tables, calculators, comparison layouts and dashboards on the fly, built on Antigravity and Gemini 3.5 Flash. Generative UI rolls out globally this summer, free for all users. Mini apps come in the following months for subscribers. Search can now generate the thing the page exists to provide.
This is the one to worry about for any client with traffic to tools, calculators, comparison pages, interactive content or quite frankly a website. Realistically, it’s a matter of adoption and cost. Will Google afford to essentially rebuild an app or website from scratch every time a user asks for one? Will people be willing to wait for it? Will I be able to sleep tonight because of the heatwave? We just don’t know!
What we do know is, people are comfortable and they don’t want to think about how an app should work or craft a prompt that will produce a reliable outcome or deal with the bugs that come with them.
How Bigwave has adapted
Generative UI changes the brief for a narrow band of pages, not the whole site. A calculator or comparison table that exists only to do a simple sum is exposed, because Search can build that itself. A calculator that sits inside real expertise and unique data, with the reasoning, the caveats and the local context that explain why the numbers matter will always be more useful to visitors than AI Search assembled from nothing.
So, the work is to stop treating tools as the destination and start treating them as the hook. We’ve moved client interactive content towards pieces that earn the visit for a reason beyond the function, things that carry the brand, the judgement and a point of view no AI can replicate. The function gets people there. The thinking around it is why they stay and why the brand gets remembered and cited, ultimately benefiting overall marketing and SEO.
The other practical change is feeds and structured data. If Search is going to build interfaces out of data, the brands with clean, accurate, well-structured product and service data are the ones it builds from. So, some of the investment moves from the front end and towards making sure the underlying data is good enough to be the source.
Personal Intelligence in AI Mode going global
Rolling out to nearly 200 countries and 98 languages, no subscription required. Once a user connects Gmail or Photos, AI Mode answers become personalised against their own context. Two users get different answers for the same query. Ranking comparisons across users stop being like-for-like, and first-party audience relationships (email lists, registered users, returning visitors) become more valuable because behavioural context shapes how a brand surfaces.
The takeaway here is that the neutral SERP (search engine ranking page) we have always benchmarked against is going away. When AI Mode can read someone’s own email and photos, the answer bends towards their context, so two people asking the same thing see different brands surfaced for different reasons. Rank tracking still matters, but it measures less of the real picture than it used to.
And as I said: “When one channel constricts, others expand!” so this just gives marketers more of a reason to get into people’s emails.
What this rewards is ‘being known’! A brand that the user already has a relationship with, through a newsletter, an account, past purchases or saved content, has a foothold that a cold competitor does not. So the work moves a little further towards earning that first-party connection and not necessarily relying on a clean ranking position on its own.
Conclusion
So is the 2026 I/O a turning point for Search? Not really. It carries on the direction Google has been moving for two years, packaged to look like a leap. The model is faster, the box is bigger, the answers are more personalised, and the interface can build itself. None of it changes the fundamentals of what works, which is being a genuine source of value that both users and the AI reading on their behalf can trust.
What it does do is keep narrowing the easy wins. Informational traffic to thin content keeps declining. Pages that exist to perform one small function get more exposed. Generic content that can be easily replicated gets absorbed into the AI answer and never earns the click. The brands that come through this are the ones with something a generated answer cannot replicate, real expertise, a clear point of view, clean data and a direct relationship with their audience.
That has been our position for a while and nothing at I/O 2026 changes it. The channels shift, the tactics adapt, the work stays the same. Be the source!