Google I/O is happening this week in Mountain View, CA. As you can imagine, this year’s product updates and unveilings are laced with AI, including Gemini, Nano-Banana, and other ways that Google’s AI efforts take shape. This is now the center of Google’s AI universe as it battles an innovator’s dilemma.
That last part is all about upending the traditional search model (10-blue links, etc.) to gain an edge in the AI race. And that sometimes means cannibalizing itself – a lesson Google learned over several years as the one doing the disrupting. It’s doing a decent job so far in being on the other end of that exchange.
All of that was on display at I/O. At the center of it all was Google’s latest AI model, Gemini 3.5 Flash. It’s optimized for agentic AI work, such as independent multitask workflows within user-defined parameters. That’s important because most of the rest of Google I/O’s updates build on this agentic capability.
For example, Google previewed the new face of search. Rolling out this summer, it will feature an “intelligent search box.” This synthesizes AI Overviews and AI Mode and makes them the primary interface, rather than a sideshow. Google calls it the biggest transformation of search in 25 years.
Information Agents
So what does this intelligent search box do? Utilizing Gemini Flash, it will use natural language search queries as a springboard to launch “information agents” to gather information on a user’s behalf. This brings search from links and answers to proactive agents that work persistently in the background.
One way Google frames this is that users will be able to build personalized mini-apps. Because we’re so app-oriented as a tech culture, this is a better way to frame AI agents. These apps can do things like search for a new apartment with particular specs, or monitor stock market activity for particular signals.
These wide-reaching and open-ended apps could be especially relevant for local search and commerce, such as alerting a user when a reservation becomes available at a new hot restaurant. Among other things, it can help users gain live market knowledge in situations that require speed and responsiveness.
Beyond search, Google is bringing these principles to other surfaces, such as Workspace and Gmail. Because Google knows a lot about you given your calendar and email, it can be well-trained as an agent on your behalf, playing the role of research assistant or executive admin. This is known as Gemini Spark.
Another manifestation of these principles is Universal Cart, an agent shopping play that can be assigned various tasks. In addition to finding time-sensitive deals – per the “live market knowledge” point above – it employs Google’s Agent Payments Protocol (AP2) to complete purchases for the more daring users.
Eye-Level
As for the latest gadgetry, Google has AR glasses coming next year and audio glasses launching this Fall. For the latter, Meta has validated a massive market with Ray-Ban Metas. Now everyone is chasing glasses that offer high-quality images & video capture, and whisper situational intelligence.
In fact, Google is calling this category “intelligent eyewear.” This naming not only reflects a reliance on Gemini, but is a plain-spoken approach that deviates from the category’s geeky tech-acronym-centric naming. This is meant to appeal to larger audiences and avoid the tech elitism that buried Google Glass.
As for Gemini’s part, intelligent eyewear puts AI at the center of the UX. It’s all about telling you about your surroundings and helping you navigate your day. In other words, it brings all of the principles above and puts them at eye level. The hardware play also brings Apple-like vertical integration to Google.
Bringing this all back to the Localogyverse, many eye-level use cases occur in the domain of local search. In fact, examples Google gave on stage included ordering and picking up coffee, and hands-free updates about your arriving Uber. Many other intelligent-eyewear user cases will develop from there.
Header image credit: Solen Feyissa on Unsplash
