In April I was in London for a few days. I needed somewhere to eat on the first night, somewhere to stay that wasn't the first result on Booking.com, and a few ideas for what to do. I didn't open TripAdvisor. I didn't search Google. I just asked an AI.
It gave me a shortlist of restaurants within walking distance that matched what I was after, a hotel with good reviews that wasn't the obvious chain, and a couple of things I wouldn't have found scrolling a top ten list. The whole thing took about two minutes.
That's not a future scenario. That's what's already happening. And last week's Google I/O 2026 conference made it clear that it's only going one direction.
Google I/O 2026 became the most-watched developer keynote in the event's history. It also had a 2% like rate on YouTube, one of the lowest engagement ratios you'll see on a video that size.
People weren't watching out of excitement. They were watching because they sensed something shifting underneath them. The technology on display was objectively impressive. The implications were, for many people, genuinely unsettling.
Both of those things can be true at once. And if you run a business with any kind of online presence, both of them matter to you.
Here's a plain-English summary of the announcements that have real implications for SMEs.
AI Mode in Search has crossed one billion monthly users. Google's AI-powered search experience, where it answers questions directly rather than just listing websites, is now the default for most users globally. The search box itself has been redesigned for the first time in 25 years, now accepting text, images, files, and videos as inputs.
Information agents are now active. These are AI systems that run continuously in the background on a user's behalf, monitoring websites, prices, availability, and news without the user having to check. You set a task once. The AI keeps watching.
Universal Cart has launched. An AI-powered shopping cart that works across merchants, tracking deals and price drops in the background. UK rollout is planned but not live yet. Worth watching if you sell products online.
Gemini Spark is Google's new "24/7 personal AI agent." It integrates across Gmail, Docs, Calendar, and Maps, summarising emails, consolidating information, and handling repetitive tasks automatically. The most convincing demo involved pulling together guest lists and RSVPs from scattered email threads into a single spreadsheet. Genuinely useful for anyone managing a lot of communication.
Gemini Omni handles any input and produces any output. Text, images, video, audio, in any combination, out in any format you ask for. Primarily a creative and developer tool for now, but the direction of travel is clear.
LLM (Large Language Model): The AI technology behind tools like ChatGPT, Google Gemini, and others. When people say "make your business visible to LLMs," they mean ensuring these systems can find, understand, and accurately represent your business when someone asks about it.
Agentic AI: AI that doesn't just answer questions but takes actions. Booking, buying, monitoring, and completing tasks autonomously. The shift from AI as a search tool to AI as an operator.
AI Overviews: The AI-generated summaries that now appear at the top of Google search results, answering questions directly instead of just listing websites. A field study in early 2026 found these reduced outbound clicks to websites by 38% on affected searches.
Zero-click search: When a user gets the answer they need directly in Google without clicking through to any website. These now account for around 72% of AI-assisted searches.
Forget the demos for a moment. Here's the practical shift happening right now.
People's search habits are changing. They're not typing keywords into Google and scrolling through links. They're asking questions in natural language, having conversations with AI, and getting answers without ever visiting a website. My trip to London in April was an early version of that. It's becoming normal.
That creates two separate challenges for any SME, and it's worth separating them clearly.
When someone asks an AI to recommend a local accountant, a good restaurant in your town, or a supplier for a specific product, does your business get mentioned? That's now a real question with a real answer, and most SMEs don't know where they stand.
Visibility to LLMs comes from the same foundations that have always mattered: a well-maintained Google Business Profile, a website with clear and specific information, consistent details across the web, and genuine reviews. Nothing exotic. But if these are out of date or thin, you're not showing up.
This is a completely separate question, and the honest answer is: slower than the hype suggests. No single person will use all of the tools announced at I/O. The right approach is to identify where AI genuinely saves you time on something you do regularly and ignore the rest for now.
Gemini Spark summarising your emails and pulling key actions is probably useful. Automating your entire customer communications pipeline in one go almost certainly isn't the right move yet.
There's a point in AI-driven personalisation where it stops feeling helpful and starts feeling uncomfortable. You've probably experienced it. A product you mentioned in passing following you around every website you visit for a week. An email that knows slightly too much about you. That unsettled feeling that someone's been watching.
Over-personalisation erodes trust. Under-personalisation feels cold and generic. The businesses that get this right, especially smaller ones, find a tone that feels like a human paying attention, not a system tracking behaviour.
That balance doesn't happen automatically. It comes from having a clear sense of who your customers are, communicating consistently, and using data to be genuinely helpful rather than just present everywhere. Getting that right is something I help clients with. The systems behind it don't need to be complex. They just need to be set up thoughtfully.
Not a checklist of everything Google announced. Just the three things worth acting on.
Sort your Google Business Profile. Complete, accurate, and updated. This is how local AI search finds you. If it's out of date, you're invisible.
Make your website specific. AI systems pull from pages that answer questions clearly and directly. Vague copy about "delivering bespoke solutions" gets skipped. Specific information about what you do, who for, and where gets picked up.
Know who your best customers are. As AI increasingly mediates discovery, having a direct relationship with your existing audience matters more, not less. That starts with a CRM that actually reflects your customer base, not just a spreadsheet of contacts.
Not everything from I/O 2026 will land. Google has a history of impressive demos that take years to arrive properly, or don't arrive at all. The Universal Cart is genuinely interesting but it's not in the UK yet. Agentic AI making purchases on your behalf is still early-stage.
What is already here is the shift in how people search for things. The move from keywords to conversations. The expectation that AI will do the filtering. The question of whether your business shows up when it should.
You don't need to implement every AI tool that got announced last week. You do need to make sure your digital foundations are solid enough that when someone asks an AI for a recommendation in your space, you're in the answer.