- New
- Insight
Postcards from Google I/O: What "AI" looks like when you stop talking about it
L+R leadership attended Google I/O 2026 in Mountain View. Here's our read on the announcements that matter, focused on where they create real competitive advantage for the businesses we build with, and where they're just noise.
A note before we begin. The two of us walking this conference are Alex and Ivan: A & I. The acronym is not lost on us. We've been a two-letter joke for over a decade, and now the rest of the world has caught up. We're here to see what the other AI is up to.
We took a Waymo to the conference
A self-driving Jaguar pulled up to our hotel. No driver, no friction, charging itself between rides. Sci-fi when we started L+R in 2012. Today, a sedan in a parking lot.
That's the thesis of this piece. The technologies that change life don't stay magical. They become infrastructure. They get boring. Boring is the highest compliment a piece of emerging technology can earn. The bar isn't whether a demo dazzles. The bar is whether, three years later, you'd notice if it disappeared.
That's the lens we'll bring to everything Google announced, read entirely through what creates real competitive advantage for the businesses we build with.
Why we're here
L+R is a studio for emerging technology, but our work isn't about emerging technology, it's about the real-world application of it. We're tech-agnostic, working across Fortune 500s, startups, and social impact, all in service of one mission: improve life with design and technology.
The easy read of a studio like ours is "if it's new, they're into it." We're not. We covered this in this year's Applied Imagination Review. Novelty alone is a terrible reason to build anything. The questions we ask are harder: Is it useful? Does it protect the humanity of the person using it? What is it doing to their privacy, their attention, their agency? The negatives in this industry are vast, and pretending otherwise is how harmful products ship at scale.
So why come to I/O? Because we're builders. The only response we have to a technology landscape this powerful is to build inside it, carefully, with intent to help. This piece is written by two people who actually have to build with what was announced.
What was announced
Google led with two numbers: 9.7 trillion tokens processed per month and 900 million Gemini app users, double last year. This is the platform a billion of your customers are already using.
The announcement we'd flag first is the quietest one: WebMCP. A new open standard that lets websites tell AI agents "here are the things you can do here" in a structured way. Today, when an agent uses your site, it's screen-scraping. WebMCP turns that into a clean handshake. The same way every brand had to think about SEO in 2005 and mobile in 2012, every brand will need to think about agent-readability in 2026. If you're building or refreshing a product right now, ignoring this is a decision, and one you'll likely regret.
The second one most clients should care about is the expansion of Google's Universal Commerce Protocol into Lodging and Food (both waitlisted). UCP is the standard that lets an AI agent actually transact for you, not just suggest a link. Today, retail. Now, hotels and restaurants. For brands in hospitality, food, and travel: integrate early or be invisible. When a user says "find me a place near the waterfront tonight," UCP-integrated brands are the ones the agent can book. The rest don't exist.
The third is hardware: the Samsung Intelligent Eyewear with Gentle Monster and Warby Parker, first revealed last year, finally got a ship date. They land this fall. Audio-only, no display, closest comparison is Echo Frames. The partnership with established eyewear brands is the first credible attempt at making smart glasses look like eyewear rather than a wearable. For brands with voice-first or hands-free use cases (accessibility, fieldwork, travel, hospitality, retail floor), there's a twelve-month window before this form factor stops being interesting and starts being expected.
The model layer underneath all of it got a step-change: Gemini 3.5 as the new flagship, with Flash at 4× faster output, and Gemini Omni as the multimodal generation model that turns text into video, image, and audio from a single call (Demis Hassabis demoed taking a selfie and editing it into a video through conversation). The companion story for builders is Google AI Studio, which now lets you prototype, deploy, and ship full agentic apps without leaving the browser. For clients, the takeaway is the same one we'd give last year, but more so: the multimodal customer experience that was a six-month, three-vendor integration is now a single API call. Brands that move first get a window.
Where the advantage Is
If we were sitting across from a client this week, three moves to make now:
- Every brand with a website: start auditing it for agent-readability. WebMCP is coming.
- Voice-first or hands-free use cases: the audio-glasses moment is now. Start designing for it this quarter.
- Product team: start prototyping with GenUI and watching Flutter. The interfaces your customers see in twelve months will not look like the interfaces they see today, and the teams that learn to design with generative tools first will ship faster than the ones that don't.
- Lodging, food, retail brands: get on the UCP waitlist.
A final note from A & I
When we say AI, we mean Applied Imagination. The industry has no shortage of imagination. What it's short on is applied: asking who this is for, what it does to them, and whether the world is better the day after it ships.
We've been doing this since 2012. The technologies that lasted were built by people who cared about the human on the other side of the screen.
If any of this applies to your business, your product, or your weird idea you can't stop thinking about, reach out.
Alex (and Ivan)
