What I Learned at Google’s Big Tech Conference, and How It Will Impact Older Adults

The agents are taking over

Search and Google’s Gemini AI app are increasingly going agentic: that is, relying on AI agents that perform tasks for you in the background, based on your instructions. These “information agents,” as Google calls them, are designed to free you up to tend to other matters on your phone or computer, and even let you close your laptop while they do your bidding. Come back later and it’s done; some tasks are shorter or longer or may require constant updates. Rivals such as Anthropic and OpenAI are also betting big on autonomous agents, which have mainly been used by companies but  should play a bigger role among consumers moving forward. 

Last year, I posed a question: Would you let an AI agent shop on your behalf? Google hopes your answer will be yes.

It introduced an “agentic commerce” Universal Cart and hub that works across merchants and services; the company claims it will give your shopping superpowers. Yes, that’s hype. But you will be able to add items to this new shopping cart you’ll see when you search, chat with Google Gemini, watch YouTube or consider a product someone mentions in a Gmail message. Once products are in the cart, AI agents proactively search for deals and price drops, and alert you when unavailable items are back in stock. Agents can also notify you if something seems amiss. For example, if you’re a tech hobbyist building your own computer, the agent may flag that the components in your cart are incompatible with one another.

This intelligent cart also knows the payment cards you have (via Google Wallet). It may recommend using one over another to earn loyalty rewards points, special discounts or other perks. Google is teaming up with a list of retailers, including Amazon, Etsy, Meta, Microsoft, Salesforce, Shopify, Target, Wayfair and Walmart, to codevelop and/or support common industry standards for payment infrastructure. But it remains to be seen how comfortable consumers will be having Google, which already knows a lot about our habits, become a kind of personal shopper.

It’s worth noting that consumers won’t need to check out with Google Pay in the cart; in some instances, they will have the option to complete the purchase at the merchant’s website.

The Universal Cart will roll out this summer across Search and the Gemini app in the U.S., with YouTube and Gmail to follow.

But is it safe?

The chief question likely on your mind, and certainly mine, is how to prevent the agent from going off the rails by paying too much for an item, or buying something you don’t want or need.

According to Google, protocols are in place to ensure that transactions are secure and comply with your directions on how much you can spend, as well as the brands, products and retailers you are willing to shop at. A digital paper trail will be captured, so if a refund is in order or disputes arise, hopefully the proper merchant will own up and pay you back.

“That’s going to be the tricky part. I’m not going to use an agent that I don’t trust,” says Carolina Milanesi, an Atlanta-based tech analyst at Creative Strategies, a market research and consulting firm. “I think we get excited about the possibilities the technology brings, but finessing the business model behind it is still going to take time.” 

Nobel Prize winner Sir Demis Hassabis, the CEO and cofounder of Google DeepMind AI research lab, is mindful of the risks and of ensuring that the agents are robust, safe and secure. “The more powerful these agents get and the more we kind of delegate tasks to them, the more we have to make sure from a cybersecurity standpoint they’re doing exactly what we expect them to do,” Hassabis said onstage at I/O. “We can monitor that.” 

Your own personal agent

Google describes Gemini Spark as a personal cloud-based AI agent or digital assistant in Google Gemini that works in the background to take complex tasks off your plate. It will integrate with apps like Gmail, Google Docs and Google Sheets, and sync across computers, iPhones and Android devices. You just tell the agent what you want it to do.

A simple example: Say you’re planning a block party and need to track not only who’s coming but what they’re bringing. Behind the scenes, Gemini Spark will kick into action and, when appropriate, generate new files. It may create a live RSVP tracker in Sheets and send follow-up email reminders in Gmail to neighbors who haven’t yet responded. If you say you want the party to have a giant bouncy house for kids to play in, Gemini Spark might warn you that homeowner association rules require you to install the house after a certain time or date, which it found in one of your Google Docs.

Gemini Spark will initially be made available as a beta to “trusted testers” and subscribers of Google’s $100-per-month AI price plan or higher.

Use your voice to create a Google Doc

For many people, writing things down and organizing their thoughts is a heavy lift. Google demonstrated a feature called Docs Live that can think everything through for you and structure a first draft.

In a real-time recorded demo, Google gave an example of a person who had to give an alumni talk at his high school. The person rambled on about needing talking points for his remarks, and told Google to include facts from his résumé and to add funny analogies to make the topic, career-day logistics, more engaging.

He also needed Google to update the event time and other details, and to put the information at the top of the document it was generating. At first, Google spit out a document the person said was too “dense.” He asked it to make some tweaks and create an easy-to-digest table, all by voice.

The whole exercise, including revisions, took barely over a minute.

Google will roll out Docs Live over the summer, but only to Google Workspace business customers and users who pay at least $20 per month for an AI subscription. A similar live-voice AI feature will also be coming to Gmail.

Tech Guru Ed Baig tests out Google’s smart glasses, which feature a small camera in the frames and speakers in the arms.

Courtesy Ed Baig

Bringing back Google Glass?

Sort of. More than a dozen years ago, Google unveiled Google Glass, a $1,500 wearable computer with a heads-up display that was famously a bust. It may have been ahead of its time, however. Iterations of very different intelligent eyewear have since emerged, most notably Meta/Ray-Ban and Oakley smart glasses, plus glasses that double as hearing aids or caption readers.

While the I/O conference is not as focused on new hardware as some other tech conferences, Google showcased AI-powered Android XR glasses in partnership with Samsung as well as eyeglass brands Warby Parker and Gentle Monster. They’re due out in the fall and will compete against Meta and others.

No pricing or product details have been announced, but Google said there will be multiple types: audio-only versions that will deliver instructions to your ears, and other models with a display.

I tried on a pair of prototype sunglasses that had a version of Gemini embedded inside. They didn’t look bad, and using my voice and hidden controls on the frames, I could snap a picture (which you can edit with the Nano Banana AI image-generating tool), find and navigate to a nearby gas station, and play Frank Sinatra. 

The smart glasses field is something to, um, keep a further eye on — including at Apple, which will host its own developer conference in June.