Amazon’s CEO Jeff Bezos Introduces Fire Phone (Full Transcript)

Prime includes fast shipping, we ship boxes to customers billions of times a year and we have to do that accurately. That takes a lot of attention to detail. Kindle Owners’ Library started with thousands of titles, and today it’s more than half a million. So we keep at it.

Let me tell you – you can fill a bucket with an eye dropper if the bucket doesn’t leak. And you can fill a bucket with a firehose but if the bucket leaks, it’s still going to be empty tomorrow. Well, Prime isn’t leaky, people use the service, they use it a lot. They love the service, and when it comes time to renew, they renew. That’s how you get tens of millions of members.

For a mental model for something like Prime, one customer – we’re not trying to build a great service for tens of millions of customers, trying to build a great service for one customer. If you think about it that way, the consequences of building a great service for one customer is you can get millions. And these individual customers do love Prime.

Amazon Prime brings so much joy to the world. Prime – what would I do without you? Seriously. I’m in a monogamous relationship with Amazon Prime. And my personal favorite – the only decent life choice I’ve ever made is Amazon Prime. [Ms. Bumcys] if you’re out there somewhere, I bet that’s not true.

So what about hardware?

Well, it’s the same thing. You have to be patient. You have to work at it. And you have to obsess over the smallest of details.

We’ve been in device business now for 10 years. We launched our first Kindle seven years ago. And we worked on it for three years before that. When we launched that first-generation Kindle, there was a lot of skepticism. We got comments like this. “The Kindle is here though it may as well be kindling”. It wasn’t hard to find those comments.

I’m extremely happy to report that over the last few years, the tone of those comments has started to change. We now get comments like this – “The Kindle Paperwhite is readers’ dream. I’m wildly impressed with the simplicity and beauty of this device”. And today we have tens of millions of Kindle owners.

The Fire HDX is a lightweight, sharp screen, superfast pleasure. I used to think the iPad was the king of tablets but not anymore. It’s time to whip the crown from Apple and give it to its rightful holder – Amazon’s Kindle Fire HDX. And today we have tens of millions of tablet owners as well.

And we just launched Fire TV. Voice search is never not amazing. You almost dare it to mess up, trying the most complicated names you can think of. By the way, never not means always. After spending some time with the Fire TV, we feel obligated to apologize for our skepticism. And we’ve been working at hardware for many years and it’s incredibly encouraging to us to see the tone of these comments. And reputation is a trailing indicator of excellence. In my view, that’s how it should be.

So let me ask this question. What’s the most important thing Amazon has done in the last 20 years? Is it developing this reputation for hardware skill? Is it building this big base of engaged Prime members? I think it’s something more foundational than that. I think the most important thing that we’ve done over the last 20 years is earn trust with customers. We’ve worked hard to do that.

Now how do you earn trust? Well, I can tell you how you don’t do it. You don’t ask for it. That never works. I think there is a simple recipe for earning trust. It’s hard to execute, hard to do but it’s simple to describe.

Here’s how you earn trust. Step one: do hard things well. Step two: repeat. Can’t do this tens of times, you can’t do this hundreds of times, you have to do this thousands and thousands of times over and over and over. But if you do that and you stick with it, customers notice.

Here is the ACSI, the customer satisfaction index. They poll tens of thousands of customers every year. Amazon is number one 4 years running. 4C? 4C does the same thing but with retail, Amazon is number one 9 years running.

The YouGov brand index – Amazon is number one, and you will see that on there, number 10 is Kindle. It’s not easy to get two brands on this list. Watch out Cheerios.

And Harris poll for corporate reputation. In 2010, we were number 9 – 2011, 2012, 2013, and 2014. Customers do notice. And one of the things — one of the hard things that customers have come to expect from Amazon is that we invent. Each of you has received a copy of this book, Mr. Pine’s Purple House. The hero of the story is Mr. Pine who wants to do things a little bit differently. My mother dutifully read this book to me at my request hundreds of times as a child. She is here in the audience today. Thank you, mom.

So we, at Amazon, like to do things little bit different – little bit differently as well. Over the last couple of years, last two years, maybe, probably the most frequent question that I have been asked – is Amazon going to build a phone? I get this question every time I meet with anybody from the media.

Sometimes the question gets asked – so when is the phone coming out? I know why people are asking this question – is Amazon going to build a phone? What they’re looking at is that we have developed hardware expertise. We have hundreds of millions of engaged customers and have this huge Prime ecosystem. They make sense to me that with looking at these skills and assets that, that will be a natural question.

But inside Amazon, we ask a different question. We ask – how would it be different? How would the phone be different? Can we build a better phone for our most engaged customers? Can we build a better phone for Amazon Prime members?

Well, I am excited to tell you that the answer is yes. Start with some of the basics.

Fire Phone – Features

This phone is gorgeous. I can’t wait for you to get your own hands on it and look at it. Pretty huge amount of effort into the industrial design. It’s elegant and refined, uses premium materials, rubberized frame, Gorilla Glass on both sides. We have CNC aluminum milled buttons. They are three times as precise as the plastic buttons that others often use.

Steel connectors, you won’t get that USB wobble that everybody hates. And we obsessed over the chamfer of the USB connectors with the [core of] the slide in easily.

Fire Phone – size

It’s 4.7 inches, absolutely gorgeous display. We’ve built models everywhere from 4.3 inches to 5.5 inches. We picked 4.7 as the perfect size for one-handed use. We obsessed over outdoor viewing – 590 nits. This is the industry-leading brightness. We put it in a circular polarizer so that even with your sunglasses on can use this phone in both portrait and landscape mode.

Under the Hood

What about under the hood? We do a lot of things with our smartphones these days. And you need a lot of power under the hood. Quad Core 2.2 GHz processor, BT coprocessor for graphics and 2 GB of RAM make sure it’s fast and fluid.

Our phones are also our primary cameras. And we put a huge amount of attention and energy into making this the best camera. 13 megapixels, a superfast 2.0 lens and optical image stabilization. That’s not what you want. But that’s what we get too often.

This is what you want. Crisp beautiful photo. Optical image stabilization, tiny little electric motors on the lens that adjust the lens 100 times a second. You’re seeing it here in super slow motion, and they counteract that natural hand tremor that all humans have. Here are some of the results. That’s unretouched straight out of the phone, with that beautiful photo. This one – same thing, gorgeous.

Let me show you a comparison. Here we’ve got Samsung on the left, iPhone on the right, and the Fire in the middle. Let’s zoom in on this. You can look at the sky, look at the sharp edges on the skyscrapers and compare it to the Samsung and iPhone. What you’re seeing here is the impact of the faster lens soaks up more light and optical image stabilization. The optical image stabilization – because it’s counteracting your hand tremors lets the shutter still open four times longer, so you can gather more light.

When you’re in bright light, great conditions, you’re going to appreciate the 13 megapixels. And when you’re in tough, challenging lighting conditions, you’re going to really love the optical image stabilization.

It’s another thing to make photography great on this phone. We gave you a button you can press, that launches the camera wherever you happen to be. You can be in the middle of an email, you can be on your lock screen, press that button, bam, the camera is up. Press it again and you’ve just taken a picture.

And we’re including unlimited storage for your photos on the Amazon Cloud. One thing about phones that often doesn’t get enough attention is the sound. We think sound is important on a phone. We added dual stereo speakers, Dolby Digital plus virtual surround. And sometimes you’re not using your speakers, you’re using your earbuds. Has anybody ever seen this? That’s what happens as the famous brilliant philosopher Bill Murray once said, “How to tie the strongest knot ever: put headphones in pocket, wait one minute”.

Well, we decided to try to fix that little bit. We got flat cables on our earbuds with magnets that class them together and – premium earbuds, we’re including them in the box. They’re not an accessory you have to purchase.

Services matter too

All right. Services matter too. It’s not just the hardware. Let’s take a look at that. Video matters. Every passing year and month, we can see this, people are using their phones more and more to watch video. It’s becoming a very important use case. And we have made video awesome on this phone. And we’re bringing some of our much loved features like Second Screen and X-Ray to the Fire Phone.

So with Second Screen, you’re watching video on your phone, can fling it to your Fire TV, your Xbox, your PlayStation, any Miracast-enabled device. X-Ray, inseam data from IMDb, so you can see every character in every scene and who the actor is, and what the tribute is and so on and so on. And we’re bringing over features to Fire Phone from Fire TV like ASAP. This is something that people absolutely love on Fire TV and it works the same way on the phone. We predict what you’re likely to want to stream and then – because we pre-buffer that, we pre-cast that for you, streaming starts instantly. People love it. And it’s an open ecosystem. You can – ESPN, HBO GO, Netflix, Hulu Plus, anything you want to watch.

What about Music?

Phones are primary way to listen to music. We’ve got tens of millions of songs you can buy. We have X-Ray for Music, so you can see synchronized lyrics. And just last week, we launched Prime Music – your library is now a million songs bigger than it was last week. And we’ve got all the apps you want there too: Spotify, Pandora, iHeartRadio Music, [N Joi].

I think you will be unsurprised to learn that we lavished attention on this phone for reading. You have the Kindle Store, millions of books to choose from, by far the largest e-book store in the world, 500,000 exclusives and the Kindle Owners’ Lending Library. Hundreds of magazines and newspapers and Comixology looks terrific on this phone.

Reading

Audible, talk about audiobooks, we like to say read while your eyes are busy, while you’re jogging, while you’re cooking, while you’re in your car. And we’re bringing much loved features Immersion Reading and Whispersync For Voice. Immersion Reading lets you read and listen simultaneously. Whispersync For Voice lets you switch back and forth seamlessly.

Listen while you’re in the car from your Fire Phone, you can get back home switch to your Kindle, read with your eyes.

Customer Support- Mayday

How about customer support? I can assure you Amazon customers care deeply about customer support. Can you roll that video please?

[Video Presentation]

So what if you need a little help? How about Mayday? We launched this on our tablets. Now more than 75% of the contacts on our tablets are coming through Mayday. People love this feature. It’s 24×7 on the phone. It works on WAN as well, works on 4G. So you can be anywhere and you can get your on-screen tech support person appearing right away. We set a goal ourselves of 15 seconds. We’ve actually been answering in less than 10. And guess what, it’s free. So you don’t have to make an appointment, drive to the store. Tap on that Mayday icon, it’s right there in the Quick Actions menu. It’s easy to find, tap on it, bam, somebody will appear on your screen, they can help you do things.

All right. How about something little bit different? I’m going to demo something we call Firefly for you. I think you’re going to like it.

Firefly

I am going to go really fast, because I want to show you how fast this is. I am going to press and hold the Firefly button and start recognizing things – Phone number, recognizes the book, recognizes the DVD, QR Code, CD, the recognizing earl, how about a game? What’s this – Kosher Salt, look at the barcode, KIND bars, Spot It, great family game, highly recommended. I have a new towel lever in my house and dishwashing detergent.

All right. So let’s pull the history on those and take a look. I did that fast as you could see how quick Firefly is. There’s that – you see the dishwashing detergent at the top and it didn’t tell, and the Spot It and so on, we can come down.

Now we recognized the phone number, we recognized an earl. I can tap on any of these items and there are actions. So I will tap on the book and you can see I can buy the Kindle Edition, I can buy the paperback. I can go into the Audiobook store. I can share with a friend and so on.

Okay. I want to also show you that Firefly can not only see but Firefly can listen as well. Can you please go ahead and play that song? Sunshine by courtyardhounds. And let’s look at the actions here. You can buy the MP3, you can buy the CD but you can also – third parties can build actions for Firefly. So here’s iHeart Radio is one example. What iHeartRadio does – they have an action that creates an iHeart Radio station using the song that I just recognized as the seed for the station. Let’s tap on that. Let’s just do it.

There it is. It’s as easy as that. How is that?

And Firefly can even recognize TV shows. Can you play that episode please?

[Video Clip]

We have just recognized the King’s Road. It’s Episode 2 from Season One of Game of Thrones. And we haven’t just recognized the episode, we’ve recognized the scene. So we are 49 minutes and one second into that scene. So now we have all the X-Ray data available. You can see all the actors in that scene. One of actions here, you can quickly just go to the IMDb page if you want more detail.

Okay. Firefly can recognize something else too. Firefly can recognize art. Let’s try this. Corn flowers, tap on the action there. We pull up the Wikipedia entry. You can learn about the artist, learn about the painting. Can you pull that expanding please? Man with Red Hat. Pull up that. There we go. Wikipedia info, learn about the painting, learn about the artist.

One more thing – you pull up the phone number – just if you’re walking down the street, you can do this from a distance too. You don’t have to be – doesn’t have to be a poster, so there I just recognized that phone number. That’s it. Super simple.

Okay. Let’s keep moving because I got a lot still to share with you. Thank you guys.

Firefly recognizes 100 million different items. And it does this under real-world circumstances. These machine vision problems are tough enough to crack in the laboratory. They’re really hard in the real world. You have things like glare, wrinkles, curves, very tough problems. And there are many techniques that we have applied to get to solve those tough problems and to get it to be this fast. I am just going to share a couple of them with you.

One of them is called semantic boosting. It’s very easy to understand. You guys know that this is the area code of a phone number. This buried in the middle is called an exchange. Well it turns out not all exchanges are valid for all area codes, and you can use that domain knowledge with semantic boosting to improve the probability that your optical character recognition (OCR) gets the phone number right. Let’s look at that.

Here is an example. You can see this is very – this is a classic real-world example. OCR often has trouble distinguishing 8s and 3s. In this particular case, it’s really tough because the 8, there is glare on top of the 8. And so when we first do this analysis, we come up with 703, which is wrong. But we know it’s wrong, because 703 is not a valid exchange for the 206 area code. So we reanalyze it excluding 703 as a possibility and you get the correct thing. This all happens totally invisibly to the user. The user never knows this and you get the right number.

We’ve done a lot of tricks too to get faster response time. You saw when I was going through that table, bam, bam, bam, how fast Firefly is.

Well here is one of the things that we do. Look at this complicated image, that’s the real image you’re going to take to your camera if you’re trying to get the phone number off of this apartment for rent sign. While we do some of this processing in the Amazon Web services cloud, where for all practical purposes we have unlimited compute power. We still have to get the image there, and a 2.1 MB image takes a little bit of time to transmit.

So we do this locally on the device. We first scrape that image to find only the relevant parts of OCRable text. You get rid of all the color and we get to compress that image from 2.1 MB to 13 kB. We reduce in size 165 times because we’re just sending the parts that matter. Makes it much faster. These are some of the  – I mean there are just tons and tons of techniques that have to be stitched together to make this work in the real world.

And all of that work on Firefly, all advanced computer science would be completely wasted if you couldn’t get to Firefly quickly? So that’s why he gave Firefly a dedicated button. Press and hold the Firefly button, no matter what you’re doing, in the middle of text message, anything you’re doing on the phone, press and hold that button even from a lock screen, even behind your pin, and bam, up comes Firefly.

The Fireflies get released and they immediately start recognizing. That’s really critical because if you have to type in your PIN or go to your app grid, find the right app, tap on something and then get it recognizing. There’s so much friction in that process, you’re not just going to use it very often. Firefly is so easy to access. Customers are going to love this.

And super exciting for us is that we’re also making available a software development kit, so third-party developers can build on top of Firefly. They can use our text recognizers, our image recognizers, our audio recognizers and our content databases, those hundred million items. And all they have to do is create a custom action. That’s what when I showed you iHeart Radio before. That’s what they did. They built a custom action that uses our audio recognizer and creates an iHeart Radio station. How long did it take them? One day.

MyFitnessPal has done the same thing. And their custom action is nutrition information. So I think it’s very useful custom action. I, however, do question the usefulness of this custom action when the example product is Chitos. Chitos are delicious but I don’t think we need any advanced computer science to tell us that they’re not nutritious. They are Day-Glo orange.

Vivino is using Firefly but they’re using the SDK at a deeper level. They’re bringing their own image recognizer, their own wine label database. And their custom actions will let you see – you can scan a wine label to tell you what the best year is, which foods pair with it, so you can operate with the Firefly software development kit at different levels. You want to use our recognizers, you can do that. If you have your own machine learning expertise in-house, you can use your own recognizers. And Firefly SDK is available immediately. We cannot wait to see how the development community is going to surprise us with Firefly. Can you please roll out TV ad?

[Video Clip]

Okay. Let’s talk about something else a little different. You see this piece of art behind me. This is what art looked like for thousands of years. It’s completely flat. And then in the 14th century, geometric perspective was invented, the vanishing point. And it changed everything. People really liked this technique. They liked it so much that it has been around for 600 years now and it hasn’t really changed. Almost every painting you see makes use of it.

Here is another one. Luxor, gorgeous painting, geometric perspective. Can you please run the video – we’re thinking maybe we can improve on a little. Can you run that video?

[Video Clip]

Dynamic Perspective

Well, what if there were 1000 artists standing by to redraw the picture every time you move your head? And what if they could do it, well, I’ll just pick a number, say, 60 times a second, here it is. We call this Dynamic Perspective.

This is one of our lock screens. We had a lot of fun making these lock screens. Show you a couple of others, let’s see. How about the forest? Falling leaves there, nice touch, brilliant designers. How about — they are brilliant. How about the lens? Gorgeous.

All right. Let me show you some other things with Dynamic Perspective. And you’re going to get a chance to play with this later. I am very excited about you getting your own hands on this. And let’s go in to maps and let’s go to the Empire State Building. Here we go. Let’s zoom in little bit more. Here is the Empire State Building.

Let’s go to our current location. There’s the current location button here in the lower left-hand corner, little black here, and we’re going to tap on that. Bam, here we are in Fremont and let’s look for some Thai food around here. Search for Thai restaurants, here they are. Plenty of Thai restaurants nearby. Let me show you Dynamic Perspective how useful is in certain situations on-screen.

So again look at – I’d draw your attention to this little current location indicator there, the black arrow. You see how it’s on a different layer from the map itself and it’s occluding the name of the street underneath it. Now as a designer on a phone, you’re often faced with this kind of trade-off. Every time you bring in on-screen elements onto something like a map, it obscures something else. And so you have to be very judicious.

But here we get to use natural movement, I can tilt just like that, it says Smith Street. Tilt to little left, I see it’s West Smith Street. It’s covered up when I am holding like this, but oh I look down, that you don’t have to teach anybody to do that. Everybody in this room, everybody in this world already knows how to do that because it’s how physical objects behave.

And we use Dynamic Perspective sensors to be able to do this. Just to tilt my phone few degrees, and up pops more information. So since it knows what angle I am holding the phone, I can see all the Yelp reviews, how many reviews there are, how many stars there are. I can do it the traditional way of swiping through. I can also just tilt a little bit like this.

Now again you would never make the default to have these — to have all this information on screen because it’s too obscuring. It obscures too much behind. But if you can just have a gentle tilt, you get that information and then tilt back and it goes away. It’s incredibly natural. That’s Dynamic Perspective.

Let me show you some other things on Dynamic Perspective. Let’s go over here, go into store. Here are some dresses. Tap on dresses. If I can do a Dynamic Perspective – so I tilt, I am controlling that browsing. How beautiful that is! If I tilt the other way, the dresses reverse, I can go back to the beginning. Those are the kinds of things that you can do.

I’ll go into this one. Like that dress, and I can use the same techniques, if I tilt just a little, I can enlarge this image. Look at the dress big, I tilt back. We call this lenticular. In some ways, it’s reminiscent of those Halloween decorations you see, but as you move your head just a little, the image changes. It’s using the Dynamic Perspective sensing system. If I tilt a little further, I can see the back of the dress. Incredibly easy to do.

Okay. Another thing that uses Dynamic Perspective – let’s go into the web browser. Here is an article. I just randomly chose a Washington Post article. I am not sure why I like that publication. Charles Wright, former UVA professor. Now I can scroll this the way you would expect to but there’s a better way to scroll on this phone because of Dynamic Perspective. I can tilt it. What I have to do is tilt this phone. If I tilt it further, it goes fast. If I tilt to the other direction, it goes backwards. Tilt it just a little.

This gesture, we call it, Auto Scroll, is so natural that now that I’ve been using this phone for several months, every time I go back to a device that doesn’t have Dynamic Perspective and can’t do this gesture, I keep trying anyway. If I keep tilting this – oh, right, it doesn’t work. You don’t have to obscure part of the screen with your finger to do that and it’s completely a very natural easy one-handed gesture. And because of the Dynamic Perspective sensors, it’s also a very robust gesture.

Okay. Let’s do the same thing or something very similar but in a book. Here is Beautiful Ruins, open that and I have the same ability – tilt the phone a little bit and start scrolling. You can tilt it further, scrolls further, I can page through this book the way you would normally do. But I can also have it scroll infinitely. It’s just one big long scroll, I go like this. And I can get the speed about where I want it. Tap and lock that in place. So now it’s just going to scroll at that speed. Yeah, again brilliant designers.

But you want to see something even cooler – let’s say, not that this ever happens in my household. I have four kids. But let’s say that I’m reading my book and somebody asked me a question, distracts me. I can put my thumb on it and stops. Lift my thumb up, stop, go. It’s incredibly natural because if there were a physical scroll moving by underneath your thumb, if you put your thumb on it, it would stop. Somebody distracts, you got to look at something, put your thumb on, take care of that whatever that interruption is and then bam, you’re back to your reading. Okay. Brilliant designers.

App Grid

Let me show you. We have our Recency Carousel, you people are familiar with from our tablets. And I am going to talk about that in a little while. But we also have a traditional app grid. It’s very useful, because it’s really good for muscle memory. Everything is exactly where you left it, so you can put things where you want them to be and you can expect them to be there when you get back.

We’ve improved on the traditional app grid, because we’ve made it possible for you to pin not just apps but content. So you can pin the book you’re reading to your app grid. You can pin a magazine, anything you want. It doesn’t have to be an app, it can also be content, pin that to your app grid.

Okay. Now let’s look at the Carousel. So here it is. And for those of you familiar with our tablet Carousel, this will be very familiar but we have made a couple of improvements. We’ve enhanced it a little bit. I am going to show you some examples.

So if I go over here to email, you’ll see that underneath – I don’t even have to launch my – I don’t have to launch the email app in order to see emails. I can scroll some. In fact, I can even do some light email maintenance. If we’re going to do a lot of email, sure, I will go into the app. But here I see this budget attached, I am just going to delete that. So that’s very useful. Having these active widgets right here underneath the Hero icons lets people dive in and do things without even opening app.

Here is Calendar and again I can just see what my appointments are. Camera, same thing. Here is my Camera Roll in most recent first order and scroll up, if I just try to get a quick picture. And third parties can do the same thing. Here is Zello. They’ve built a active widget here for nearby homes, so you can see the homes nearest in a view, ready to move to Seattle, here you go.

All right. Let me – see, I want to show you music, switch to one that’s wired up, so you can hear as well. All right. What I want to show you on music – here this is an example of showing our three panel design. Here is the Justin Timberlake song and I personally – and I am really extremely sincere about this. This is not a joke. I have unbelievable admiration for anyone who can make a tiny fedora cool. That is – I mean, really, I am being sincere. That’s incredible. It’s harder than building a smartphone.

Three Panel Design

Now so everywhere throughout this phone, we have a three-panel design. So the left panel is usually navigation. So here you can see playlists, artists, albums and so on. Then there’s the center panel, which is kind of the heart of it. And in the right panel, there’s always some kind of lighter – in this case, a synchronized lyrics. So let’ do that. Playing the song. They are the synchronized lyrics right there. I haven’t even talked about it, but you see how natural that gesture is. All I am doing is tilting the phone and these panels shift in and out. We’ve made that gesture very robust. Lyrics, you can also find a particular place in the song. Skips right through it, there you go.

Let’s pause that. All right. Now I am going to show you – a game. I think you can probably imagine what the Dynamic Perspective technology can enable for gamers, and I am just going to show one very quick thing here. I am going to show you a game called Tofu Fury. I picked this one because just the idea of Angry Tofu delights me. It’s just fantastic. Go in, and what you can do here is look around on this image. So I’m tilting the phone to look everywhere at this level. So this – well, I will kind of outline what you’re trying to do in this game very briefly.

I can maneuver this block of Tofu and my challenge is to collect all these blue orbs. That’s the Chi. And after I’ve collected the Chi, or as much of it as I can while avoiding the deadly spikes. You see those deadly spikes. Then I have to save fortune Kitty – the pink fortune Kitty there.

Okay. So this game is actually really fun. I am going to – I can hop like this. I have to hop high enough to get the blue orbs but low enough to not have – we’re going to call that good enough. But now you notice, I can look around, tilt phone and that’s a very natural gesture. So from this I look up, I tilt to the right and I see fortune Kitty over there, let’s go. Yes, bam. Okay, save fortune Kitty.

All right, good. That is Dynamic Perspective.

So when we show this to people for the first time, the first question they ask is – how the heck you even do that? Well, the key is knowing where the user’s head is at all times. We have to be able to know where the user’s head is at all times, in real time, many many times a second.

We started working on this – actually all of these things – four years ago, little more than four years ago now. And we had early prototypes of Dynamic Perspective working in the first week. And some of the things — if you got to know where the user’s head is at all times, you can do this. You have a piece of headgear with infrared lights on it, it’s really easy to track the user’s head then. The one problem with this – the user has to wear this piece of headgear. The right way -– and that would never be commercially practical. The right way to do this is with computer vision.

Now that’s a tough problem. It’s a tough problem, even the lab, super-hard when you move out into the real world. In the real-world, you got all these problems. You got really bad lighting, lots of variety, all kinds of problems when you’re trying to track somebody’s head robustly. If you want to do this as a prototype, you can get something started. But if you want to do it for commercial product, you have to do it robustly.

Now it’s very tempting at the beginning. In fact, we prototyped this as well. Just say, well, we’ve already got a front facing camera on the phone. Let’s use that existing front facing camera to track the head. Well, it turns out that’s not going to work. The whole bunch of reasons. One of them is that the field view on the front facing camera is too narrow. We needed a widering of field of view, you can see here, if you use this field of view, it’s very typical. The user’s head goes either part way or all the way out of frame too frequently as you’re using the phone in a very natural way and it wouldn’t be robust.

So we made special cameras that have a much wider field of view. And now that’s better. Under a lot of movement, a phone stays – I mean face stays in the frame. Okay. So that’s one problem solved – wider field of view, specialized camera.

That’s not the only problem. You need to know the Z-depth, you need to know how far away the head is, which means, you need stereo vision, which means you need two cameras. So we added two cameras – two of these specialized cameras.

Great. Now you have good field of view, got two cameras, you got stereo vision, you’re making progress but you’re not done – because users hold their phones in a whole bunch of ways and they end up obscuring the cameras. Oops… now we’re back to mono vision.

So we added two more cameras. Now I have got four cornered cameras. And no matter how you hold your phone, two of them, we can pick the best two, we’ve got great stereo vision.

But there is still another problem. People use their phones in absolute darkness. They lay in their bedroom at night, end up using their phones in dark cars. So these cameras also had to work — have to work in pitch black.

Well, we solved that too. We added infrared lights, one for each camera. You can’t see them. They’re invisible to humans. But they work great for the cameras. And we made these cameras in a very special way. They’re global shutter cameras instead of rolling shutter cameras. And as a result, they are 10 times as power efficient and then in software we also turn them on just when we need them, at just the right times.

So with all of this hardware development, we end up with this – great stereo images in any lighting conditions. That is a very tough thing to achieve. But it’s just the beginning – because everybody is different. People wear sunglasses, they wear sunglasses on their foreheads. They have beards. They have hats. They have big hair. They have no hair, that’s my personal favorite. And they even have almost invisible eyebrows. All of this variety is extremely difficult machine learning problem.

And if you want to solved really tough machine learning problem, you know what you need? Lots of data. You need tons and tons of data. In this case, images, in order to train your algorithms. So we started out in the laboratory.

Over here you can see a Fire Phone being manipulated by one robot. And on the left, there’s another robot manipulating that Mannequin head and yes, I agree that that’s a little creepy looking, yes. And then the disco lighting is to create really harsh difficult lighting conditions and collect images in all of those and we can change that head. We can do a lot.

And you can get a long way with this kind of laboratory setup. You can collect millions of images and we can create very challenging environments for ourselves with shadows and lighting and so on. But ultimately, to get a robust product, you need to move out into the real world. And we did that.

We collected millions of images with the actual camera hardware from all over the world. By the way here I am compressing years of work by some of the world’s best computer scientists into two minutes, for which I apologize with that T. At the end of all of this, we got really good at tracking faces, finding heads. Think about the complex background, see that like the dock or whatever that is over the third guy’s head there. Very difficult, very challenging.

Okay. So now we’re really good at tracking faces, finding heads, doing that in real-time. We’re almost there. We’re not quite done, because once you get really good at finding heads, you can find them in a lot of places we don’t want to find them. Here is one example. This is a real example. This mug belongs to one of our developers. Here’s what happened. Good, on the faces – oh, no, not the cherub. Yeah, you don’t want the cherub.

Now it turns out this happens with T-shirts that have two bog on them everything. And we have found the heuristics and workarounds for all of these things. This one – there is an ingenious solution to this, to the cherub problem. And it’s ingenious and incredibly simple. Remember we know the Z-depth. So we know how tall that head is. That cherub head is an inch and a half tall. No human heads are that small. We can reject that head.

Here is another example. Very difficult, extremely challenging. We solved about one too.

So at the end of all that hard work, here is what you get. Three numbers – X, Y and Z. You can see that blue dot. It’s showing you the X and Y Cartesian and the size of the blue circle around it showing you the Z-depth. And once we get those three numbers in a robust way, we then tie them in to our 3-D renderer and you get this.

Same thing with the auto scroll that I showed you. Tilting the phone just a little bit to control in a very fine-grained reliable way, the speed of scrolling with one hand. Same exact thing. Instead of sending that XYZ data to a 3-D renderer, we’re sending it to the auto scroll code. And the Luxor screensaver that I showed you. X, Y and Z – 3 numbers.

This is so beautiful and elegant that all of that hard work, years of work, blows down to these three numbers. And one of the reasons that that is so great, that it’s so simple, the output is so simple is that it makes it easy for developers to take advantage of this.

We’re excited to announce the Dynamic Perspective software development kit (SDK)available immediately. Developers are going to blow us away with what they figured out how to do here. Can we run the video please?

[Video Presentation]

Okay. Developers are going to love these SDKs.

I am going to show you another thing here. Small touches. Now in a launch event like this, we tend to focus on the big things. But some of these tiny little small touches that you add various places on the phone can make a huge difference. Things that people do all the time and they don’t get all the attention. But when users really use the phone, they often appreciate these small touches.

Okay. Let me show you a couple of things. Let’s go into the calendar. Here is one. So here is my calendar. I look at my calendar. I realize I am going to be late to my next meeting. I can go to the right panel and it says, go ahead and start without me, running a couple minutes late. Be there in 10 minutes. Sorry, we’ll have to reschedule. So I can send a message to all the people that are already invited to that meeting and say, go ahead and start without me… form as the message, I press in, I am done. Incredibly easy. Right there from the right panel.

Another one – here is – I hit the volume button to pull up this. I got my ringer on, I can switch it to vibrate, I got the ringer on, I can switch it to silent. But what if you’re going to go to a movie, or you’re in a meeting, and it’s just going to turn off silent for three hours and then it counts down, and at the end of the three hours, it turns your ringer back on. Do you know how many marriages this could save? It’s the small touches – one people are going to love.

Let’s – another one. Here is something. We were looking, remember, for those things that people do frequently and wished were easier. And sometimes they get so inured to the old way of doing it. They don’t even know that it could be easier.

Let’s go into messaging. One of the things that people do very frequently is send photos with messaging. Let’s look at our right panel. Bam, these are all my camera photos in most recent order. So if I just took a photo, it’s going to be right at the top. I took it a few minutes ago, maybe down here. Here is a fun face. Tap that, goes right in, press send, I am done. I encourage you to duplicate that, do the same thing on your current smartphone and see how many taps it takes you.

All right. One last one, let me go to the lock screen. Okay. Here is that lock screen. And these lock screens are so gorgeous that people are not – you can replace it with your custom photo, your choosing. We think many people are going to want to not do that, because lock screens are so cool. And so we wanted to give them something else that would be even better, so they could have their cake and eat it too, or eat their cake and have it too, I don’t know.

Anyway with this, got the beautiful lock screen, let me do a swipe, and I have a whole photo album. I can choose a photo album that I want to be one swipe away from my lock screen. So this can be your kids, can be your family, it can be your last family vacation, whatever you want, there is a whole photo album here. Gorgeous, simple.

Okay. So those are some small touches. There are dozens of small touches like that throughout the phone, and I’ve just highlighted a few to sort of give you a taste of it.

I’m very excited to tell you that the Fire Phone is going to be offered exclusively on AT&T. We’ve been working with AT&T for five years now with Kindle. They’ve always done an amazing job for us, and they’ve been great partners with the Fire Phone. We could not have asked for better.

I want to introduce Ralph de la Vega, the CEO of AT&T Mobility. I don’t know where – here he is. Come on up, Ralph.

Ralph de la Vega – CEO, AT&T Mobility

Thank you. Thank you, Jeff. First of all, thanks for the opportunity to join this great event. And thank you, Jeff for the opportunity to collaborate with Amazon once again on what I think is some amazing breakthrough innovation. Don’t you guys think so? Wow.

I’m really impressed by what Amazon has done with Fire. It’s effortless to get to applications, to navigate websites, identify objects, it is compulsive once you learn how to do it. But it’s also a great phone on its own.

What I really think makes the phone different, though, just like Mr. Pine’s Purple House are four key features of the phone that, as I have used them, and I have used a lot of phones, are absolutely breakthrough and innovative. And the first one is what he just described Dynamic Perspective. Those four front-facing cameras provide a picture that is absolutely amazing. When I have shown that phone to people, I get one response. And it’s wow! And it’s consistently wow! They have never seen anything like that.

So imagine what’s going to happen when software developers get hold of those SDKs and figure out how to do even greater things with this technology. It is wow right now and I think it’s going to get a whole lot better.

The second feature that I really, really love and I hope you got a great picture of it was Firefly. I’ve had Firefly in my office and I was running around the office Fireflying everything. I actually went into the kitchen, I went into the pantry, I Firefly-ed everything and they came out bing, bing, bing, with 100 million objects, the chances are it’s going to work.

Now I am also an Amazon Prime member. I’ve used Amazon forever. So I was showing my assistant all the stuff in the pantry. She says, “Let me hold it”. I said, “You can’t touch it. If you touched it, you distorted it, [it would lock] the technology. It was just amazing. I think people are going to look at phones in the future and wonder why they don’t have this capability. I am telling you it is addictive and it is an absolute breakthrough that we’re so glad you’re bringing to market. I am going to buy a whole lot more things with this technology than I ever have before.

The third thing – and this is a subtle thing but I was equally wow-ed by Fire TV. And when you put the combination of Fire Phone and Fire TV, what you get is amazing. What I did – I was able to not only turn up the Fire TV almost instantly. I have 45-max service in my home. It’s great, delivered by AT&T by the way. But I also have the entire house with Wi-Fi. I took Fire TV out of the box and I just hooked it to power into my HDMI connection. And that’s all I needed. Everything came in through Wi-Fi, the screen came up, it had my information and I actually looked at a three minute video that explained how to use it.

My wife was sitting next to me. I started using the voice recognition feature and just like Jeff told you, I’ve got everything right. That never happened to me before. You know, I have a Spanish accent, that’s really hard to do. So I tried to fool it – I tried to give it my strongest Spanish accent I could and it still got it right. I gave it to my wife. She tried it. They were right. I was just really blown away and the fact that when you walk in the house, if you’re watching instant video on your phone, you can slink it to your Fire TV and then use your phone as a secondary screen to get all the information about that movie or that picture, it’s just amazing. And that is AT&T’s vision that, that smartphone is in essence the remote control for your life. You can control your TV. You can control your house. You can control your car and anything in between, and what you have seen today is Amazon bringing that reality to bear right here today, not years from now but today. It’s amazing thing.

Now if it turns out, the last thing that I love about this service is – if you have trouble, then you can use what Jeff explained Mayday. It comes up easily, it’s fantastic but I wanted to tell you that we have worked closely with Amazon so that if you have any issue with an AT&T service, we have a warm handoff between Mayday and AT&T’s technical support, so you can actually get transparent and instant collaboration between Amazon and AT&T. If you have an issue with this phone or any of our services, you’re going to get it solved in a very, very cool way which I think is the way we’re going to do it for tomorrow.

Now I know you’re wondering right now, okay, I’ve got this. When is it going to be available? Well, I can tell you that beginning today, you can start pre-ordering for the Amazon Fire Phone today. And you can get this phone with a traditional two-year contract or with AT&T’s Next program on AT&T.com or at retail stores beginning today.

Now when you get it with Next – I want to highlight this. When you get it with Next, first of all, it’s zero [dime] and you get the best service rate plans that AT&T has to offer. And there’s no activation fee. There’s no upgrade fee, and no annual contract. It’s going to be a great deal.

So in closing – yeah I’m impressed. I hope you have been impressed, because I have seen Jeff and the Amazon team develop this technology and continue to develop, [Don Fuller] was absolutely right. So I commend you for doing the hard things right. But I believe that Fire and Amazon are redefining the customer experience with what they have. And we’re very proud to power the Amazon Fire with 4G LTE, so we can deliver an amazing experience to all of you .Thank you, Jeff. It’s been a great honor working with you. Look forward to making some great things happen.

Jeff Bezos – Founder and CEO, Amazon.com

Thank you very much. And we wanted you to have this. This is just the very first Fire Phone off the line. We wanted you to have it.

Ralph de la Vega – CEO, AT&T Mobility

Thank you very much. Hurray, ooohh, thank you.

Jeff Bezos – Founder and CEO, Amazon.com

So what you get with the Fire Phone – gorgeous, industrial design, premium materials, fit and finish, obsessive fit and finish, the new Dynamic Perspective sensors and Dynamic Perspective technology. Firefly button recognizes 100 million different things. Our much loved Mayday is going to be awesome on the phone. Unlimited photo storage, dozens of small touches, Amazon’s exclusive features like X-Ray, ASAP, Amazon’s huge content ecosystem, and AT&T’s incredibly reliable 4G network. We’re super happy to have AT&T as our exclusive partner. We really appreciate all the work they’ve put in to make this happen.

And Fire is going to be $199 with a two-year contract – or the $27 a month with the Next program that Ralph was just talking about. And it’s not – these were the prices you might expect for a 16 GB model. We’re making our model 32 GB for the same price. You can pre-order today and it ships July 25. You won’t have to wait long.

One more touch. This one isn’t that small. We’re also kicking things off with 12 months of Amazon Prime included. If you’re new to Prime, you will have 12 months, and if you’re an existing Prime member, we’re going to give you 12 months anyway. Can we run the video please?

[Video Presentation]

And I hope that you guys have as much fun using this phone as we had building it. Thank you so much for coming.

Related Posts

• Why eSIMs Are the Smart Choice for Mobile Businesses
• Transcript: Elon Musk and Jensen Huang Talk AI at U.S.-Saudi Investment Forum – 11/19/25
• The Minds of Modern AI: Jensen Huang, Geoffrey Hinton, Yann LeCun & the AI Vision of the Future (Transcript)
• Transcript: NVIDIA CEO Jensen Huang’s Keynote At GTC 2025
• Mo Gawdat: How to Stay Human in the Age of AI @ Dragonfly Summit (Transcript)