Google Launches Agentic Shopping Tools Ahead of Holidays

Bloomberg

Watch Google Launches Agentic Shopping Tools Ahead of Holidays - Bloomberg on Dailymotion

Transcript

00:00This is interesting because we're going from the domain of agentic AI and we're trying to take it

00:05to how consumers already use search, marry them in the context of retail. How does it work and

00:12why is it important? Yeah, so agentic tech is, you know, relatively new and we're really focusing

00:17on bringing agentic AI for consumer problems that really matter to consumers. And so, for example,

00:22taking the tedious parts of shopping and having agentic AI, which is AI that does things on behalf

00:27of the user. So, for example, with price tracking, you know, we have the scenario of you see a product,

00:33maybe the price is too high. You can track the price and we will auto buy the product for you.

00:37On the other hand, for example, let's say you have a particular product. Let's say Labubu is one of

00:42the products of the season and you want to find it near you. It will actually make all the calling

00:46for you for nearby businesses and tell you whether it's in stock. Do you have any evidence or data

00:51that this technology change results in an actual purchase or more that increases the likelihood

00:58that the consumer goes through with it? I mean, anecdotally, we do see that consumers are really

01:04loving this in the sense that both consumers and merchants, actually, especially for the agentic

01:08calling, because it's allowing consumers to have more confidence, let's say, that the product is

01:12actually in store. There are some that perhaps don't want others' agents coming and making purchases.

01:18They want that direct relationship. We've seen reports of Amazon doing a cease and desist

01:23on perplexity because of payments being made. How do you think you'll navigate the ecosystem

01:29in that way, Lillian? Yeah, it's a great question. You know, I think all of this agentic technology

01:35is very nascent. And again, at Google, we're really starting with what are those real problems

01:40that consumers have and always giving consumers control of the experience. So in the agentic checkout,

01:46for example, as I mentioned, you know, we're starting with a real user need, right, which is

01:50that a lot of times people will see a product, maybe it's too expensive. We let you track the

01:55price of the product. We alert you when the product has reached that price. And then you can auto buy,

02:00essentially, have Google buy that product for you on your behalf. And in this case, for the merchant,

02:05it's good in the sense that, you know, maybe they would have lost that customer otherwise,

02:09right? Maybe they weren't ready to buy it at that price. And so we think this is good for merchants

02:13and also for consumers. On the back end, Lillian, we've got a beautifully geeky audience. What is it

02:20that you've managed to do in the last few months, years that has just allowed for this that is now

02:26seeing the agents able to run, run free and make purchases on one's behalf safely?

02:33Yeah, I think, again, so a lot of this is new, I will say that, you know, on the agentic calling

02:37piece, we have been doing things like duplex, which you may or may not be familiar with,

02:42which is where we're using AI to essentially call a business on your behalf for some years,

02:47actually, you know, since 2017, 2018, I think. For restaurants, you've been doing it for me for a

02:52long time. Yeah, exactly. On the calling piece, I think we've been doing this for a long time. And

02:56we're really finding real user needs, you know, for for tedious parts of shopping where we think it

03:02can help. On the agentic calling piece, also tracking the price is something that we've done over

03:07some time. This year, we've actually gotten down to the variant level. So for example, if you see a

03:12shoe that you like, it really matters the size and the color of the shoe. And so now we're able to

03:17agentically, you know, really track that price. And then the buying really what we're doing there is

03:23we are instantiating a browser and really adding a product to the cart on the merchant site and

03:28letting you buy it. The question is always what happens next? And how is this going to go? At the

03:33enterprise scale, the issue with agentic AI is that you need many agents to be able to communicate

03:38with one another. In fact, in the consumer, think about an airline app, for example, and they maybe

03:43use a different software platform to book the flight originally. Are you encountering those problems,

03:49you know, Gemini and what's outside of Google's control? You know, I would say that it's still

03:53nascent. So today, for example, a lot of our merchants and the way that they show up is actually

04:00through relationships that we've developed over the years. So we have something called the

04:03shopping graph, which is, you know, 50 billion product listings of all these merchants around

04:07the world. But as you say, over time, you know, potentially you'll imagine a world where

04:12a merchant has an agent and we will be interacting between the Gemini agent and that agent as well.

04:17Yeah.

Category

Transcript

Be the first to comment

Recommended