Google I/O 2026 wrapped up, delivering a lot of news for us to digest. From a reimagined Search box to agentic commerce and a staggering infrastructure commitment, Google laid out a vision that touches every corner of digital marketing. Here’s a breakdown of what actually matters, structured around the questions marketers and businesses are asking right now.
What’s changing about Google Search? – Intelligent Search Box, AI Overviews, AI Mode
AI Mode is not (yet) becoming the default in Google Search. That’s short-term relief for publishers and brands, but it’s not a strategic reprieve. Across dozens of individual announcements, Google rolled out an agentic Search architecture that will become visible over the summer and autumn.
As Liz Reid, VP of Search at Google, put it: “Google Search is AI Search through and through.”
The headline launch is the new Intelligent Search box – described as the biggest upgrade to the Search box in over 25 years. It now accepts text, images, files, videos, and even Chrome tabs as input. AI-powered suggestions go well beyond traditional autocomplete, and the box dynamically expands to accommodate longer, more natural-language queries.
There’s also a seamless handoff from AI Overviews to AI Mode: ask a follow-up question from an AI Overview and it flows directly into a conversational AI Mode session. This is live globally on desktop and mobile.
Perhaps most telling are the official US insights Google shared:
- The average AI Mode query is 3x longer than a traditional Search query
- More than 1 in 6 US searches now use voice or images
- Image searches are growing over 40 per cent month-over-month
- Planning queries are growing 80 per cent faster than AI Mode queries overall
These numbers paint a clear picture: users are shifting from short keywords to long, natural-language prompts – and Google is building the infrastructure to match.
In short: everything is changing beneath the surface, but not the default experience.
What are Information Agents and why should marketers care?
Google introduced Information Agents – background AI agents that monitor the web 24/7 against user-defined criteria. Think of them as an AI-powered, always-on version of Google Alerts, but far more intelligent.
A user could, for example, brain-dump their flat-hunting requirements and the agent would continuously scan for matching listings. Or a sneaker enthusiast could set an agent to notify them the moment a favourite athlete’s collab drops.
Why this matters for marketers: freshness becomes an agentic ranking factor. These agents will continuously crawl the web for relevance and change, which means content needs to be kept up to date, not just published and forgotten. If a brand’s product data is stale or incomplete, it simply won’t surface in these ongoing agent-driven scans.
Information Agents roll out this summer, initially for Google AI Pro and Ultra subscribers.
What is Generative UI in Search? – Gemini & Antigravity
Powered by Gemini 3.5 Flash and Antigravity, Google Search can now build custom layouts, tools, and mini apps on the fly – directly within the search results page.
Imagine a student asking how two objects interact near a black hole: Gemini builds a custom interactive simulation right in Search, complete with links to research papers. Or picture a family planning a weekend trip: Search proactively suggests building a custom planner, connects to Gmail and Photos, and creates a shareable app in about two seconds.
Jordan Koene, CEO of Previsible, framed the strategic shift well: “Our calculators and tools will now be native to Google Search.” The implication? Brands need to shift from owning the interface to becoming the trusted data source and authority that powers these AI-generated experiences.
Generative UI rolls out this summer, free for everyone.
What are the new Gemini models and why do they matter?
Gemini 3.5 Flash is now the default model globally. The numbers are impressive:
- 4x faster than other frontier models (12x in Antigravity)
- Beats Gemini 3.1 Pro across all agentic benchmarks
- Less than half the price of comparable models
Gemini 3.5 Pro is coming in June, and Gemini Omni – a multimodal world model announced by Demis Hassabis – is live today across the Gemini app, Flow, and YouTube Shorts.
For marketers, the practical takeaway is speed and cost. These models power everything from AI Overviews to Generative UI to Information Agents. Faster, cheaper, more capable models mean Google can roll out AI features at scale. And that rollout is accelerating.
What is Gemini Spark and how does it work?
Gemini Spark is Google’s new 24/7 personal agent. It runs on dedicated Google Cloud virtual machines, meaning it keeps working even when you close your laptop.
During the demo, Josh Woodward showed Spark compiling information across email, docs, and chat; building event spreadsheets and slide decks; and even connecting with Instacart to order snacks when it spotted “snack duty” on a calendar.
Spark is powered by Gemini 3.5 and the Antigravity 2.0 harness, and it supports third-party tools via MCP (Model Context Protocol) in the coming weeks.
Rollout: beta to Google AI Ultra subscribers in the US next week, with Spark in Chrome and on macOS coming later this summer.
How is Google changing online shopping?
This is where things get particularly interesting for e-commerce brands. Google announced three interconnected shopping features:
Universal Cart is a cross-merchant checkout system that works across Search, Gemini, YouTube, and Gmail. Users can add items to their cart while browsing Search, chatting with Gemini, watching YouTube, or reading Gmail. The cart proactively finds deals, tracks price drops, flags product incompatibilities, and surfaces hidden loyalty points.
Launch partners include Nike, Sephora, Target, Ulta Beauty, Walmart, Wayfair, Fenty, and Steve Madden. Rolling out in the US this summer.
Universal Commerce Protocol (UCP) is expanding to Canada and Australia in the coming months, with the UK to follow. New verticals include hotel booking and local food delivery. Google’s Vidhya Srinivasan described UCP as doing “for agentic commerce what HTTP did for the web.”
Agent Payments Protocol (AP2) enables agents to make secure payments on a user’s behalf, with strict guardrails: specific brands, products, and spending limits. Its first deployment will be through Gemini Spark.
A crucial reassurance for brands: “The brand stays the merchant of record” (Vidhya Srinivasan). The customer relationship at the point of transaction is preserved.
What does Google’s infrastructure investment signal?
Google’s 2026 capex sits at 180 to 190 billion USD – roughly 6x what it spent in 2022. That’s not a typo.
The new TPU generation includes TPU 8t for training (nearly 3x the raw computing power of the previous generation, distributed across over 1 million TPUs globally) and TPU 8i for inference. Both deliver up to 2x better performance per watt.
The scale of this investment signals that Google is not experimenting with AI in Search – it’s committing to it at a level that’s difficult for competitors to match.
What should marketers and SEO professionals do now?
Three strategic shifts stand out:
-
Content needs to become agent-readable
Structured, fresh, easy to extract, easy to cite. Information Agents will monitor the web continuously, and freshness is becoming an agentic ranking factor. The era of publishing a surface-level blog post to rank for a keyword and collect easy clicks is over.
As Dan Taylor put it: “SEO isn’t dead, but the game is changing and we need to change with it.”
-
Technical SEO gets a new layer
Several new standards and tools emerged from the developer keynote:
- WebMCP is in origin trial – a structured surface for agent interaction
- HTML-in-Canvas API makes DOM elements accessible to agents
- Lighthouse now verifies llms.txt for agents (not for Search)
- Aria labels and the accessibility tree are becoming agentic ranking factors
Dan Taylor’s framing is spot-on: “Most agents navigate the web using the accessibility tree. Every Aria role or label that you optimise doesn’t just help a human; it makes your site more actionable for an agent too.”
-
E-commerce optimisation becomes agent optimisation
Product data needs to be complete, accurate, and structured enough for AI systems to make purchasing decisions. Brand preference will increasingly be shaped before the user ever visits the shop. Universal Cart and AP2 mean that if a brand’s product information isn’t agent-ready, it risks being left out of the conversation entirely.
The bottom line
Google I/O 2026 was the coordinated launch of an agentic Search architecture. As Johannes Beus of SISTRIX observed: “Google rarely changes Search in one step. The change runs in many small, continuous adjustments. And that’s exactly where Google remains unbeatable.”
AI Mode hasn’t become the default yet. But the pattern is the point. The brands and marketers who start adapting their content, technical infrastructure, and product data for an agent-driven world now will be the ones best positioned when these features reach full scale.
The question isn’t whether Search is changing – it’s whether your strategy is changing with it.
Want fresh insights delivered to your inbox?
Subscribe to our newsletter!