Google I/O 2026, translated for small business owners
Google's I/O 2026 keynote was three hours of AI. Most of it won't touch a small business this week. But three things will: dramatically cheaper capable models, AI Mode search crossing a billion users, and agentic shopping. Here's the signal, and what to do about it.
Google I/O 2026 (the keynote was May 20) was, like every keynote now, almost entirely about AI. Most of it is either consumer features or enterprise tooling that won’t touch a small business for a year. But a few announcements change the calculus for the kind of companies we build for. Here’s the translation, minus the keynote adjectives.
1. Capable AI got cheap: this is the big one
The headline model was Gemini 3.5 Flash. The interesting part isn’t the benchmark bragging (it beats last year’s Pro model on coding tasks); it’s the pricing. Google is positioning it as doing serious “agentic” work (multi-step tasks that used to take a flagship model) at less than half the cost of comparable frontier models, and it’s already the default model behind Google’s AI search.
Why a small business should care: the per-use cost of adding a genuinely useful AI feature to your website, app, or internal tools just dropped again. A support assistant, a quoting helper, a document-summarizer: the things that were “maybe worth it” at last year’s prices are now comfortably worth it. We’ve made this argument before about local models; the cloud side just moved the same direction.
The caveat: cheaper inference doesn’t make a bad AI feature good. The cost was never really the blocker for most small businesses; figuring out a use case that saves time was. More on that below.
2. AI Mode search passed a billion users, and your SEO problem changed
Google said its AI Mode search (the conversational, answer-first results) now has over a billion monthly users, runs on Gemini 3.5 Flash by default, and that those queries are roughly doubling every quarter.
This is the announcement with the most direct impact on a local business, and it has nothing to do with adopting AI yourself. It’s about being found. When a potential customer asks Google “who does kitchen renovations in Etobicoke” and gets a synthesized answer instead of ten blue links, the question becomes: does the answer mention you?
Showing up in those answers rewards the same fundamentals as classic SEO, just more strictly:
- Structured, machine-readable content (clear headings, real answers to real questions, schema markup).
- Accurate, consistent business information across your site and listings.
- A fast, accessible site the crawler can parse.
The classic local-SEO fundamentals still hold, but the bar is rising. If your site is a slow, JavaScript-heavy template that’s thin on actual content, AI search will quietly route around you.
3. Agentic shopping is coming to checkout
Google previewed a Universal Cart (buying across Search and the Gemini app) and “information agents” that complete tasks for users, rolling out over summer 2026. The direction is clear: people will increasingly ask an assistant to find and buy rather than browsing storefronts themselves.
For a small e-commerce or booking business, this is a “watch closely, don’t panic” item. It’s US-first and early. But the takeaway rhymes with the SEO point: your product data, prices, and availability need to be clean, structured, and accessible to machines, not just pretty to humans. If you run a storefront, the structured-data work we do is the same work that makes you legible to these agents.
What’s not urgent
To be clear, most of I/O 2026 is not something to act on:
- Gemini Omni, Google Pics, voice editing in Docs: neat, consumer/creator features, mostly gated behind paid Google AI Pro/Ultra plans and rolling out over the summer in the US. Nice to have; not a business decision.
- The $100/month AI Ultra plan: aimed at heavy individual users and developers, not small teams.
- Antigravity 2.0 and agent SDKs: developer infrastructure. Relevant to us when we build for you; not something you need to evaluate.
If a vendor cold-emails you next week insisting you need to “adopt Gemini Omni for your business,” that’s a sales pitch, not a strategy.
What to do
- Pressure-test your search presence. Ask Google’s AI Mode (and ChatGPT) the questions your customers would ask. Are you in the answer? If not, that’s the highest-ROI fix, and it’s not an AI project; it’s content, structure, and speed.
- Pick one AI use case with a real time-saving, not a feature you bolt on because it’s 2026. The cheap-model news means it’ll cost less than it would have; it still has to earn its place.
- Make sure your site is machine-readable: structured data, fast INP, accessible markup. It’s now doing double duty: ranking for people and for the assistants answering on their behalf.
That’s the whole signal from three hours of keynote. If you want help separating the one useful AI feature for your business from the 99 that aren’t, that’s a conversation we have all the time.