As search shifts from Google to AI like Gemini, DXDasia shows how to future-proof your website so it stays visible in AI-driven results.
Quick answer: Google Search moved from blue links to AI Overviews (May 2024), AI Mode (US launch May 2025) and Gemini-powered agents (2026). Websites now win by being indexable, citable and recognised as an entity: content in the server-rendered HTML, structured data that matches the page, accurate business listings and brand mentions across the web.
In three years Google has gone from a page of ten blue links to an assistant that writes answers, runs dozens of searches on your behalf and, increasingly, books the table or buys the product itself. Each step was announced on Google's own blog with a date attached, so the direction of travel is not guesswork. This article lays out that timeline, explains what each change means for a business website, and sets out the roadmap we use with clients for the next 12 to 24 months.
Google moved in four stages: an experiment (SGE, 2023), AI summaries for everyone (AI Overviews, 2024), a conversational AI search tab (AI Mode, 2025) and agents that act for the user (2026). The foundation was laid much earlier, in 2012, when Google started modelling the world as entities rather than keywords.
When Google launched the Knowledge Graph on 16 May 2012 it described the goal as understanding "things, not strings", starting with more than 500 million objects and 3.5 billion facts. Every AI feature since sits on top of that idea. The table below lists the milestones that matter most for website owners, each from Google's own announcement unless noted.
| Date | Milestone | What was announced |
|---|---|---|
| 10 May 2023 | Search Generative Experience | SGE launched as a Search Labs experiment in the US, in English |
| 8 Feb 2024 | Bard becomes Gemini | "Bard will now simply be called Gemini" |
| 14 May 2024 | AI Overviews | Rolled out to everyone in the US, with a target of over a billion people by the end of 2024 |
| 5 Mar 2025 | AI Mode (Labs) | AI Mode introduced on a custom Gemini 2.0, using a "query fan-out" technique |
| 20 May 2025 | AI Mode for all US users | AI Mode rolled out in the US; AI Overviews had over 1.5 billion users in 200 countries and territories |
| 21 Aug 2025 | First agentic features | Restaurant reservations across OpenTable, Resy, Tock and others; AI Mode extended to over 180 new countries in English |
| 7 Oct 2025 | Global AI Mode | More than 35 new languages, over 200 countries and territories |
| 18 Nov 2025 | Gemini 3 in Search | First Gemini model in Search "on day one", with more fan-out searches and generative layouts |
| 11 Jan 2026 | Universal Commerce Protocol | Open standard for agent shopping and checkout in AI Mode and Gemini (TechCrunch) |
| 19 May 2026 | Search agents | AI Mode passed one billion monthly users; Gemini 3.5 Flash became the default; agents and agentic calling announced |
| 3 Jun 2026 | Controls for site owners | Search Console AI reporting and an opt-out toggle, rolled out worldwide by 31 August 2026 |
| 22 Jul 2026 | One AI search experience | AI Overviews and AI Mode brought together, sending "billions of clicks to websites every week" |
The scale is the point. By June 2026 Google said AI Overviews had over 2.5 billion monthly active users, and in August 2026 the Gemini app passed 1 billion monthly users. AI answers are no longer a feature some searchers see. They are the default way Google responds.
Four things change: how your page is retrieved, how often people click, whether the assistant can act on your information, and how much your brand's wider reputation counts. Each needs a different response.
Google's documentation says both AI Overviews and AI Mode may use "query fan-out", issuing multiple related searches across subtopics to build one response. Deep Search, announced in May 2025, can "issue hundreds of searches" for a single question, and Gemini 3 was built to perform "even more" searches. Queries are also getting longer: Google reported that people in AI Mode ask questions nearly three times longer than traditional searches.
In practice, a page is no longer matched to one keyword. It is retrieved because it answers one of the hidden sub-questions well. Pages that state facts plainly, under headings that mirror real questions, give the system more to pick up.
Independent studies show a clear drop in clicks. Pew Research tracked 900 US adults in March 2025 and found users clicked a result in 8% of visits when an AI summary appeared, versus 15% without one, and clicked a link inside the summary in just 1% of visits. Ahrefs found that an AI Overview correlated with a 34.5% lower click-through rate for the top result in April 2025, and with a 58% lower rate in its December 2025 data.
Google tells a different story. It says clicks from AI Overview pages "are higher quality, where users are more likely to spend more time on the site", and in July 2026 it said AI features send "billions of clicks" to websites each week. Both can be true: total queries are rising while the share that ends in a click falls. The safe planning assumption is fewer, better-qualified visits.
Fewer clicks make every remaining conversion count, so the tracking has to be right. Across several client accounts we have found the same pattern: a website redesign or relaunch removed the Google Ads conversion tag, and the account kept spending while reporting zero or wrong conversions. Re-verify conversion tags after every rebuild before judging what AI search is doing to your results.
Since August 2025 AI Mode has been able to find restaurant availability across booking platforms. At I/O 2026 Google extended agentic booking to more local services in the US and introduced the ability to "ask Google to call businesses on your behalf". For retail, the Universal Commerce Protocol lets shoppers check out from US retailers directly inside AI Mode and the Gemini app, and it was co-developed with Shopify, Etsy, Wayfair, Target and Walmart.
An agent can only act on what it can read reliably: your hours, phone number, prices, stock and booking links. Google's own advice for multimodal and AI search includes keeping Merchant Center and Business Profile information up to date. A wrong opening time used to cost a visit. With agents, it costs the booking.
Ahrefs analysed 75,000 brands in December 2025 and found that YouTube mentions (about 0.737) and branded web mentions (0.656 for AI Overviews, 0.709 for AI Mode) correlated with AI visibility far more strongly than backlinks, which showed very weak correlations. Correlation is not causation, as Ahrefs notes, but the pattern fits how these systems work: they recommend entities they recognise and see discussed.
On your own site, Organization structured data "can help Google better understand your organization's administrative details and disambiguate your organization". It has no required properties, so add the ones that apply: name, logo, address, telephone, contact points and sameAs links to your official profiles.
No. Google states that to appear as a supporting link, "a page must be indexed and eligible to be shown in Google Search with a snippet", and that there is "no special schema.org structured data that you need to add" and no need for new AI text files. The foundations are the same technical requirements as classic Search.
What Google does ask is that structured data matches the content visible on the page, that pages return a 200 status and are not blocked, and that you use preview controls such as nosnippet and max-snippet deliberately, because "more restrictive permissions will limit how your content is featured" in AI formats. Since 31 August 2026 every site can also see AI impressions by page and country in Search Console, or opt out entirely. Google says opted-out sites receive no traffic from generative features, but the choice is not used as a ranking signal for other results.
For a fuller breakdown of how AEO, SEO and GEO differ, see our guide to AEO vs SEO vs GEO, and for on-page tactics, how to make your website visible in Google AI Overviews.
The most common reason a site is invisible to AI search is not a missing tactic. It is a basic access problem that classic SEO tolerated. Three examples from our own site in September and October 2026:
None of these is specific to Gemini, which is exactly the lesson. Google says page experience and crawlable, indexable content apply to AI results as much as to blue links. Fix access first, then optimise for citation.
Work in three phases: make the site fully readable now, build entity and content strength over the next two quarters, then prepare for agents that transact. The order matters, because each phase depends on the one before.
If you want this done as an engagement, it is part of our SEO and AI search service, and the speed work sits under performance optimisation.
Most "future-proofing" advice chases the newest feature. The evidence points the other way. Google has said repeatedly that there is no special markup for AI search, eligibility still starts with being indexed, and the independent data shows clicks falling while brand recognition matters more. So we recommend spending the first quarter on unglamorous access fixes, the next two on becoming a clearly defined, widely mentioned entity with genuinely useful pages, and only then on agent integrations. A business that does those in order will be ready for whatever Google ships at I/O 2027, because every release so far has rewarded the same foundations.
On most phones, yes. In March 2025 Google said it was upgrading mobile users from Google Assistant to Gemini over the following months, with the classic Assistant no longer available on most mobile devices later that year, and the change extending to tablets, cars, headphones and home devices. Devices that cannot run Gemini keep the older Assistant for now.
Yes. Since 31 August 2026 every website can use a Search Console toggle to opt out of appearing in and grounding Google's generative AI features. Google says opted-out sites receive no traffic or impressions from those features, but the choice is not used as a ranking signal for other search results. Preview controls such as nosnippet and max-snippet also limit what can be shown.
Yes. Google's documentation says traffic from AI Overviews and AI Mode is counted in the overall Web search type in the Performance report. Since mid-2026 Search Console also shows separate insights for generative AI features, including impressions, which pages appear in AI responses and in which countries. Use both to see whether AI visibility is growing even if clicks are flat.
Google says you do not need one. Its AI features documentation states there is no need for new machine-readable files, AI text files or special schema to appear in AI Overviews or AI Mode. Some other AI services may read such files, so treat llms.txt as optional and low priority, well after fixing crawlability, server-rendered content and accurate structured data.
Query fan-out is Google's technique of splitting one question into several related searches across subtopics and data sources, running them at the same time, and combining the results into one answer. Deep Search applies it at a larger scale, with hundreds of searches. For site owners it means a page can be cited for answering one sub-question well, even if it does not rank for the original query.