Instacart’s Smart Cart Uses AI to Watch Your Every Move | Fast Five Shorts

Instacart is integrating Nvidia AI technology into caper carts, that smart cart platform.

Turning each cart into an AI edge device and laying the groundwork for what they call the grocery world model.

Now, Grocery world model, what is it you might ask?

Oh yes, please tell me, Carter.

Yeah, tell me more about this.

I want to know more.

Well, according to Chainstorage, each caper cart is now going to be equipped with the Nvidia Jetson Edge AI device that runs advanced sensor fusion.

So sensor fusion buzzword, right?

Combining basket facing cameras, weights measures and a certified scale location tracking system for when it's actually in the aisles and outward facing cameras to accurately identify the exact items that are placed in the cart in real time, even when the cameras are blocked.

Talk about a surveillance device.

Holy cow.

So two AI systems are going to work in parallel.

The Edge encoder, which Nvidia Jetson is known for, giving real time feedback and a cloud vision language model for all the reasoning and actually crunching of that data.

Those two data streams produce really this signal that allows for this incredibly high accuracy to understand everything that is in the cart.

The system updates its understanding of each store shelf as frequently as every hour, creating a real near real time view or model of shelf conditions, item placements, etc.

Right Chris, this is the A and M put you on the spot question of the week.

And here it is.

As investment continues for the caper cart is the Nvidia integration and specifically the idea that I can see items on the shelf unlock Instacart needs to increase capers relevance or is throwing more money and hype at this a niche application?

Oh my God.

It's 100% the latter, Carter.

100%.

This is, this is the, this is the hype of all hype announcements.

And I mean they're laying AI on top of, you know.

Yeah, the announcements maybe get a little bit more press.

The amount of ways this card has been tried to be sold over the last X number of years is just unfathomable to me.

So like this, here's the.

But here's the reason I think this, this announcement is all hype, because to me this is this.

Actually the funny thing about this announcement is it's right strategy, wrong substrate.

For example, if the cart is the tool for this, why does it need to be a smart shopping cart at all?

Couldn't you just equip a regular old cart unbeknownst to shoppers and not require shoppers to adapt their behavior to using a new cart and get all the same information that you're reportedly trying to get from this, or 100% you can.

And with that said, though, there's even a better way to do this.

And that better way is robotics.

Because the problem with the cart serving as the substrate is that despite what this headline says, you cannot guarantee the real time view of the store is being captured the same way at the same time every time.

Because when you're leaving that to customers or employees, they're going to do it differently.

A robot is the only thing that can do it the same way every time.

And robotics are still the best way to do that because that's what they do.

And their field of vision is better too, for that reason as well, because they can get inside the nooks and crannies better than, say, a cart that's just randomly going through the store.

So once again, once again, Carter, like Instacart's out pushing the cart.

I don't think it's going to work.

They'd be better off at this point.

They'd be better off just acquiring a robotics company and stop throwing more good money after bad.

That's my opinion.

Yeah, it's so the other thing to think about here also, first of all, there's like multiple downsides, different angles here.

First of all, the consumer side, they talk in this article, Chris, about, like, what is the benefit to the consumer?

Like, why would I take, you know, two minutes to log in and actually do this thing when my kids are crying, right?

And I'm just trying to get in and out, whatever, you know, and they say, oh, it'll help you with recipes.

Well, coming from someone who did digital at General Mills, like, no, like this recipe game.

And the recipe excuse is something that everyone talks about, but no one actually can figure out how to add any value to.

You got your lists that you put on your phone or written on a piece of paper, and that's the behavior, that's what you do.

Don't try to change it.

And the value that they claim this is going to give the consumer pales in comparison to the friction that it's going to bring, which is accurate.

Then you have the actual tech side.

So, Chris, one of these Nvidia chips is about 500 bucks.

That doesn't account the cameras, it doesn't count the scales, it doesn't account the actual cost of the cart.

Right.

How many shopping carts have you seen in a creek down by your house?

Right?

Like the cost that this is going to bring to grocers, Right?

Like, you know, you're talking about a retail environment that's got 2% margins and you're going to spend, what, $2,000 per cart so that you can get a little bit more data between the hours of noon and one, that coffee is out of stock.

Like, it just doesn't, it doesn't make sense.

That relationship is so off balance.

There's zero way that this actually is a legitimate play.

Yeah, that's really interesting, too.

Like, actually, that's a good, that's a good tell for all our listeners, too.

When you, when you hear the stories that people keep saying over and over again, in this case, it's recipes.

Like, yeah, I mean, you basically just said on the previous headline, like, you couldn't even sell recipes through TikTok.

You know, it wasn't, it wasn't really doing anything for you through TikTok.

So it's not going to do anything for you on a cart either.

You know, it's so, it's just, it's just as I think about.

And I say that too, because when I'm sitting here at Shop Talk, you know, agentic AI is the topic of the day.

I'm going to veer off course a little bit, but agentic AI is the topic of the day.

And you keep hearing all these things about how it's going to make our lives easier.

We don't have to reorder everything, which is the same baloney that you heard about subscription models forever and ever.

And it's just because our minds just gravitate towards something we know without just going into and saying, okay, how many of us really want that at the end of the day?

Because the chances are it's a really small percentage of the people.

I don't know.

What do you think, Carter?

Am I, am I on the right track with that?

100%?

Like, here's the thing, like, there's two sides.

You have the industry side, which everyone's just talking about buzzwords, and I guarantee no one actually knows what it means or how to actually apply it to their business.

Right.

That's the issue that you're seeing at Shop Talk now and then from the consumer perspective, like, they just don't care.

Right.

Like, you know, again, I go through my mind of the barriers to entry to try to assemble a recipe on a touchscreen that your kid is yelling at you about.

Right.

Like, it just is the worst.

Right.

So it's just one of those things where, you know, there are, there are fails in every side.

This is a sizzle, as we used to call it, Chris.

This is a sizzle in a pan.

I was going to bring that up, actually.

Yeah.

This is throwing AI at a problem.

Chasing bad money after more bad money after more bad money and thinking that AI is going to be the solution with these really expensive carts, when the investment probably could have been made into something like robotics, like you talk about.

Right?

Yeah.

100% Something that's proven and rolling out.

Oh, my gosh.

Showing the efficiencies that come with it, too.

That's the other thing.

Like, why would you be investing this?

It's totally unproven, but anyway.