Amazon AI Update: New "Feeling" Robot, New AI Coder
09:23 Share

Today on the podcast, we are talking all things Amazon. So there's a couple of interesting stories. The first one is that Amazon is currently working on AI code generator. I think this is particularly interesting when you take into context the fact that Amazon owns AWS, Amazon web servers, where you could say the majority of the web's code is hosted.

running, I guess. So there's a lot of, I think, interesting plays happening there. And the second thing is that Amazon, and this one was kind of a shocking headline to me and maybe a lot of other people.

Amazon has shown off a robot in their warehouse, which has a sense of touch. So we're going to be getting into what that means, what that looks like, what their code thing is. Amazon's doing a lot of exciting stuff with robotics and AI right now. So we're getting into all of that. Before we do, I wanted to mention that AI Box...

The AI Box Playground is officially launched and in beta. And if you want to try it out, it is essentially a tool that I have been developing for the last two and a half years. We have more exciting stuff coming out of AI Box in the future. This is our first product. And so I'm super excited. It is $20 a month and you get access to all of the top AI models. So Anthropic, Cohere, DeepSeek, Google, Meta, Microsoft, Mistral, NVIDIA, OpenAI, CorelDRAW,

quen grok and then that's just for the text models for the image models we have black forest labs which is flex it's kind of what grok uses as well there's ideogram which i've been incredibly impressed with opening eyes models and of course we have text to speech models speech to text models um so audio and the cool thing with ai box is that you have the ability to talk to all of these models text audio and um

Image all in the same chat is switch between different models. They take into account the context and you can Essentially determine what models are best you can compare models side by side their outputs a whole bunch of exciting stuff You don't have to pay subscriptions for all 20 of these models You can pay 20 bucks a month and get access to all of them. So you check it out on AI box AI I am so excited to have this officially launched. All right, let's get into what Amazon's doing

So the first thing I wanted to talk about is their warehouse robot. This thing is so fascinating. Essentially, they're calling this thing Vulcan. And they say it can, quote, feel some of the items that it's touching. So it has two arms. It's a robot with two arms. It can maneuver inside of their warehouse. And it goes essentially to the storage compartments. And it uses what are called force sensors to help it know when it makes contact with an object. So one of the arms...

on the robot is essentially rearranging items into a compartment. And the second arm has a camera and also suction cups and it goes and grabs the item. What absolutely blew me away about this is that Vulcan right now, it was trained on physical data, which includes like force and touch feedback, and it can pick up around 75% of Amazon's stock.

Which is impressive. And it's also capable of self-improving over time. So a couple of big things here. Number one, Amazon's building this robot. It's got AI embedded in it. It can go around and about 75% of what's in an Amazon warehouse, this robot can grab. So there's probably 25% that might need specialty robots or humans or other people, you know, forklifts or other things to do.

This robot can get 75% of it and it's going to get better. What's interesting is it's capable of self-improving over time. The cool thing with AI is that when you have this embedded in there, you essentially are telling the robot like, okay, what did you struggle to pick up? What were you unable to grab? What were you unable to do? And then...

It's essentially learning and teaching itself how to do some of these tasks it wasn't able to do before. So it's getting better and better. And it's not because someone wrote a new algorithm or wrote some new code and deployed it that it made it better necessarily, although I'm sure they're working on that kind of stuff. It's able to learn and teach itself, which is mind-boggling and very, very cool.

So they've actually deployed this in two different places. They have it in Spokane, Washington, and then also in Hamburg, Germany. And apparently it has processed half a million orders so far. So obviously this is a test. It's a prototype. They're putting this out there, but half a million orders, 75% of all the products they ship it's able to do. So this is definitely, I think, one of the most impressive new robots. Amazon's been doing robots in the warehouse for quite a while.

It uses hundreds of thousands of robots today to fulfill customer orders. And they're doing this, you know, all across the globe and a bunch of different storage facilities. So a lot of people that are critical of this are saying, hey, look, Amazon's just, you know, replacing humans and there's going to be less human workers. Amazon is saying, look, like there's robots in a warehouse where there's dangerous things going on. There's forklifts, there's things moving around. This

This is just making our warehouses safer for humans. These probably aren't tasks that humans should have been doing anyways, if we could avoid it. And so now we just have robots doing this. I know that people are going to like be critical and on both sides of that argument. I, for one, will say, have a good friend who's, um,

who knows someone that worked in a warehouse and was killed by a forklift a couple years ago. And it's a complete tragedy. You know, I think there's definitely ways we can try to avoid this and we can work on, you know, workplace safety and all sorts of things. But at the end of the day, when you're working with heavy machinery, when things are moving around, a lot of these environments are subject to dangerous, you know, situations. And so I think when we have robots that are capable of doing things that could be dangerous or harmful for humans, I

I am all on board with that. It doesn't necessarily need to be a human who's sweating it out, breaking their back to pick up heavy boxes. This is a great job for a robot, in my opinion, my humble opinion. Any case, let's talk about what Amazon is doing with code. I think this is also an absolutely fascinating story.

So the new tool is called Kiro. Business Insider was reporting on this. And this isn't like a big official announcement from Amazon. It's just, you know, Business Insider is like, hey, we saw some internal documents. Of course, we know that's not always 100% accurate. But I don't think it's very far off because currently Amazon has another tool called QDeveloper. It's pretty much like GitHub Copilot. And they have...

have that already an AI power kind of coding assistant. So this is kind of like their next generation. What's interesting here is you can essentially use prompts and then any existing data in your data set to create code. It says that it does it in near real time, which is kind of interesting, meaning I'm assuming like you tell it to do something and instantly it's like done. I just think that talks to the latency and how fast it's able to do.

It does this by connecting with AI agents apparently. So apparently it is on web, it also has a desktop app and it has multi-modal capabilities, right? So it's able to see and hear and kind of have and look at text and images and all that kind of stuff.

This is super, super fascinating in my opinion, and it can be configured to work with third party AI agents. So not just one, but many, which is kind of interesting, right? You could imagine some of these AI agents might be tasked to go do something or go write code, and then they might actually use Kiro to go and execute that, which I think is fascinating and a very interesting thought. Like AI agents might not be able to figure out how to literally do every single task in the world, but they can figure out how to plug into an API or a system

that can get that task accomplished. In my opinion, this is where a lot of the AI agents will move in the future, as there's very specialized tasks they might not be able to do. So Kiro could be a very interesting thing. It could do a lot of things that other coders or other code places can do, like creating technical design documents. It can flag potential issues and it can optimize code, all that kind of stuff.

But to me, the most exciting thing is the fact that it ties in with all of these AI agents. So this is an incredibly popular field right now. If you look at the numbers cursor, which is essentially

owned by a company called AnySphere, has raised close to $9 billion. Sorry, it has close to a $9 billion valuation. And its big competitor is called Windsurf. And that is apparently close to being acquired by OpenAI for $3 billion. So there is a lot of money. This is an incredibly valuable asset in the AI space. These AI code tools help a ton. I, for one, am using

Over at AI Box, we're using Claude Code, which is kind of in research preview, so it's not like an official launch product. But it is writing hundreds of lines of hundreds of thousands of lines of code for us, rewriting, optimizing,

stuff that we were absolutely blown away by. So these tools are incredibly powerful and Amazon is jumping in a full force with this, with their new Kiro product. So very, very cool. Hey, if you are interested in getting access to all of the different AI models, including, you know, anthropic, which is famously hosted and has partnerships with AWS. So a lot of the stuff Amazon's working on, you can get all of that over at the AI box playground, my own software, where you pay $20 a month and get access to all of the top AI models over

Over 30 models are there right now that are all the top models that I personally was paying subscriptions for many, most of them. And now we get them all for just 20 bucks a month, all in the same place. You could chat with them all in the same conversation. It's amazing. Hope that you try it out. Let me know if there's any exciting features you'd like to add. We're rapidly developing on that. Thank you so much for tuning into the podcast today and I will catch you next time.