The Changelog: Software Development, Open Source - Werner Vogels predicts the future (Interview)

Episode Date: December 4, 2025

Amazon CTO, Werner Vogels, stops by to help us explore his tech predictions for 2026 and beyond. Will companionship be redefined by consumer robots? Will quantum-safe become the only safe worth talking about? Is this the dawn of the renaissance developer? We're infinitely curious why Werner came to this particular set of conclusions. Are you?

Welcome, everyone. I'm Jared and you are listening to The Change Log. Where each week, Adam and I interview the hackers, the leaders, and the innovators of the software world. We pick their brains, we learn from their failures, we get inspired by their accomplishments, and we have a lot of fun along the way. On this episode, Amazon CTO, Warner, Vogels stops by to help us explore his tech predictions for 2026 and beyond. Will companionship be redefined by consumer robots? Will quantum safe become the only safe worth talking about? Is this the dawn of the Renaissance developer? We're infinitely curious why Warner came to this particular set

of conclusions, are you? But first a big thank you to our partners at fly.io, the public cloud built for developers who ship. We love fly. might too. Learn all about it at fly.io. Okay, Warner Vogels predicts the future on the change log. Let's do it. Well, friends, agentic postgres is here. And it's from our friends over at Tiger Data. This is the very first database built for agents and is built to let you build faster. You know, a fun side note is 80% of Claude was built with AI. Over a year ago, 25% of Google's code was

AI generated. It's safe to say that now it's probably close to 100%. Most people I talk to, most developers I talk to right now, almost all their code is being generated. That's a different world. Here's the deal. Agents are the new developers. They don't click. They don't scroll. They call. They retrieve.

They parallelize. They plug in your infrastructure to places you need to perform, but your database is probably still thinking about humans only because that's kind of where Postgres is at. Tiger Data's philosophy is that when you're agents need to spin up sandboxes, run migrations, query huge volumes, a vector and text data, well, normal Postgres, it might choke. And so they fix that. Here's where we're at right now.

Agentic Postgres delivers these three big leaps. Native search and retrieval, instant zero copy forks and MCP server plus your CLI plus a cool free tier. Now, if this is intriguing at all, head over to tigerdata.com, install the CLI, just three commands, spin up an agentic Postgres service, and let your agents work at the speed they expect, not the speed of the old way. The new way, a Genti Postgres, it's built for agents, is designed to elevate your developer experience

and build the next big thing. Again, go to tigerdata.com to learn more. When we have a treat, we're here with the CTO of Amazon. My gosh, talking about predictions every year you have five. They mostly come somewhat true in a couple years. So you're pretty accurate, Werner. Welcome to the show again. The last time you're here for the no-sequel Smackdown back about 12 years ago, Jared, 10 years ago, what was it?

Forever ago, episode 18. So back on NoSQL's cool, fresh and burgeoning. Now it's ubiquitous and everywhere and I guess still cool, coolish, coolish. Yeah, engineering has come a wrong way since. I think the first NoSQL tools were really the first tools. Now, if you look at DynamoDB, if you look at Mongo, these are super robust, highly scalable tools. where everybody can build a business home. Come a long way.

I think, you know, it was Burjian. It was Dynamo DB was coming out then. It was fresh and brand new. I guess as a CTO, you probably architect a lot of that stuff, right? What do you do as part of your role? Like, what is your role maybe then 15 years ago and then how is it now? So first of all, it's 21 years now, right? So that's a bit.

And there is a predecessor to DynamoDB, that's Dynamo. And one of the things, when I joined Amazon, and I joined us what was called a director of systems research. The idea was to bring you know, let me go a little bit back. Think about 1994.

When Jeff Basil starts thinking about sort of the internet and he starts a bookshop. He doesn't really want to start a bookshop. He's just fascinated by the internet. What are the things that you can? do online that you will never be able to do in real life. And he just picks a bookshop. A good bookshop has 40,000 titles in stock, yet there's millions of books out there.

So he thinks you could do that online. But there is no book. There is no software that you can buy. There is no book that says, and here is an e-commerce operation. The word e-commerce doesn't exist yet. So everything that the Amazon engineers had to do to build Amazon was to invent everything themselves. Because the kind of technology that they could buy

couldn't operate at their scale. We've had a number of bloody noses because of that. So when I'm joined Amazon, engineers at Amazon were brilliant at scaling. But for a, let's say, practical, lots of scars kind of approach. Oh, yeah? And Jeff hoped that by bringing a former academic in, you get some more robustness. You know, you get some more, a better fundamental approach to scale and reliability and things like that.

And we did lots of large projects around removing all single points of failures or how to best measure, how do you measure? What does it mean to measure? Yeah, if 50% latency on your webpage means nothing. Yeah, well, it means that 50% of your customers are getting a worse experience. You need to know how much works. Now, so, and I think if I think about CTOs, There's sort of four or five different types.

And there's no real well-developed, well-described one. I think, now, first of all, there's the data center manager. He reports up to the CIO. They have the CTO that is the second person in the startup. After the co-founder, the first coder. And then you have the big thinker, sort of the role that I got when I came into Amazon, sort of like look at the whole picture. what other kind of things we need to do.

But then, when we started doing AWAS, your world changes. You go from an internal focused CTO to an external focused CTO. And I think Scott Dietzen, who was at B.A, I think was that, at that time we called it, yeah, he called it really an external technologist. The ability to talk to your customers, look at that, how are they using my products?

And what are the problems that I see with 10, 20, 100 of my customers that they may not see as a single problem themselves, but where you can find, if I can build a tool for that, I can really help my customers. And so your role changes from being purely internal to being. external and sort of bringing back things into. And then then over time, I've become more and more interested in those organizations, both profit and non-profit, that try to solve hard problems. And with hard problems, I mean hard human problems. The United Nations expects that by 2050, have two billion more people.

I mean, how are we going to feed them? How are we going to make sure they have an economic future? How are going to make sure they have health care? Those kind of problems are the ones that I'm mostly focused on today. On one hand, by the way that we build technology at Amazon, but also by looking for those often young businesses and how can we support them? Take, for example, the Ocean Cleanup Project.

It's a massive problem. The ocean, the Grand Ocean Garbage Patch is full of fish nets and plastics and there's about 30 rivers that sort of contribute mostly to that. So these guys have built plastic with GPS in them, threw them in the river and see them where they end up. Or they have these AI cameras on boats to sort of, first, where do these boats go? But secondly, also, what are the things that they're seeing? And so, working with those kind of customers is extremely satisfying

because we're dissolving actually real human problems. We're not building spam filters. Well, 21 years is a long time. And there's that old quote, the best way to predict the future is to create it. You've spent 21 years with your teams creating in many ways the future that we're currently living in. But you're also out there predicting it. This is the fifth year perhaps. You're writing these predictions. You have five of them. And they're fascinating. I agree with most of them. The companionship one I'm skeptical of, but I'd love to hear your full case for us here. But why? Why write these? Why predict the future in these ways and continue to do so? There's a long history by large technology companies. And if you think about the IBMs and the protocols and whatever, to sort of, or McKinsey or whatever, the consulting companies,

by kind of predicting the future. And always when I read them, I always felt that they were a bit self-serving. And the things that I saw in the real world, who are different. I see significant human problems, you know, and some of them are our own nature, but quite a few are caused by technology or the way that we use technology.

There's a leadership principle of Amazon. With success and scale comes broad responsibility. That means you don't only have a responsibility towards your shareholders to make money. You also need to serve your customers, but at AWS, of course, that is builders. How can we help builders best? And then, you know, I write these predictions about things I see around me. I have the fortune to travel the world.

You know, so I was just in, or just, in Spain, I was in substantive. have Africa. So Nigeria, Rwanda, Kenya. Now, in Kenya, we were in Nairobi, and of course, you stay in the center of Nairobi, beautiful buildings, whatever. You drive 15 minutes and you see how the real Kenyans live. And if you think how they work, most of them are day laborers. That means that they go to the bank, they get $2, they try, they buy something, they try to sell it, they bring the $2 back and hope to have 40 cents left to buy food. They probably have that, but then they don't have money left to cook it. Why?

Because these big canisters, these butane canisters, cost $10. And they don't have $10. And so there's a young company called Koko Networks that build a sort of ATM where you can go with a canister, you plop the canister, and give me 15 cents of gas. And then you can take that home. You can cook your food. Now, this is not a world-shocking problem,

but it is solving really hard problems that people have. And those kind of companies are really the ones that, and those kind of problems are often kind of the things that I like to service through the predictions. this year I do think there are things every year I mean I could have done

10 predictions yeah you could have but that doesn't work that fairly well five is the magic number three is a magic number five's better

yeah but you know was it two years ago an example a topic that has become more and more openly discussable

role is menopause for women. It's a major problem for them, you know, but even mothers don't talk to their daughters about it. However, that is slowly changing. What kind of things do we see in the startup community that are trying to build, whether it's technology products or whether it is medicine or whatever, to solve those kind of problems? And so, I also try to talk a little bit about things that are people normally find a bit harder to talk about. Well, let's talk about one, companionship, loneliness, robots. I do not have any disagreement whatsoever with your casting of the problem, which is that loneliness is on the rise, especially amongst elderly.

But even across all demographics, we see an increase in loneliness, which has all kinds of health and mental wellness problems, just overall bad, right? But you seem optimistic because we have a new swath of robotic companions that are coming out. More like it's one of these problems. Well, you see the problem. Actually, I did this documentary series called Now Go Build TV series. Probably two, three years ago, I was in Japan for that. Actually, for the, and that's where I first really dove deep in this particular topic.

So Japan really was a society where kids take care of the elderly. The grandparents live with the grandkids and the kids. And that was normal. But the younger kids want to start making career. There is a clear shift in Japanese society happening where the elderly are no longer taken care of by the younger people. such, you know, that's a shocker. Not for the young kids. It is a shocker for the old people. They're really on their own. And I think, and indeed, I think that there are a lot of technology

solutions possible to help these people. One of the, one of the things, and loneliness is something we'll rebuild, we'll get to, but anything we can do as technologists to keep people longer out of hospitals or whatever, care homes, or whatever, the better it is. Yeah, and if that means that, you know, a company in Japan that I met called ZWorks, they try to do all these innovations around these elderly people to help them, right, make a make a in your mattress a little bit of a sensitivity pad. And maybe in the middle of the night, you go to the toilet and you don't come back within 15 minutes.

The alarm goes off. Yeah. And this doesn't seem like world shocking kind of revolutions. But it does mean that people can stay independent longer. They can stay. Yeah. And they don't. feel that old. Let's put it like that. Well, I think more and more people, I mean,

becoming a hundred years old is actually not that strange anymore, is it? Yeah, and getting to that particular age, many of these people may have lost their partner if they had a partner. or live far away from their from other family i mean especially in the u.s i mean after high school you probably move away from where he lived where he grew up yeah you go to a college somewhere and upstate new york you go to work in california and people are not close anymore that truly is a loneliness epidemic. So what can we do?

And I think, and I love the work that is happening around sort of companion robots. Yeah. Sort of, why? Because I've, I always thought that people would treat a robot like a piece of metal

mechanics, you know? Right. Turns out it's not with all the kings. They treat them like pets. Really, Kate Darling, MIT, she's done all the research around here, is I can tell you absolutely amazing stories about how people get attached to their devices. Yeah, and the story I think is also in the predictions, 80% of the people that have a room bar, you know, the cleaner in the house, have given it the name. And one of the stories that Roomba tells, like, well, one sent an old Roomba back because it needed to be repaired and they look at it and they tell the customer, you know what, we'll just give you a new one. And they say, no, no, no, no, no, no. We want mineral fruit back.

Really? Yeah? Yeah, I want the old one because they're attached to it. Now, or Kate tells this story where at some moment she divides two groups of people. people. And each get a dinosaur, a sort of baby dinosaur and they will fall in off if they give it a name, they play with it, blah, blah, blah. And then she gives him a hammer and she tells them, kill the dinosaur. Yeah. Nobody in the group will take the hammer and kill the dinosaur. They have no problem though going to the other group and kill their dinosaur. You know?

But so now, this is sort of, why do we tell these stories is that while these things are technology, we treat them as living beings. And especially, I've always, of course, because before I was in computer science, I was in healthcare, you know, the application of these kind of tools, stories in hospitals are incredible. You know, if you're a young kid, a doctor is someone that is tall

and, I don't know, dangerous or whatever you want to think. But for example, there's this, the huggerball robot is a small green thing with a nice snout and talks and talks to the kid and whatever. Kids with a hugger ball,

have no problem talking to a doctor. Yeah, because they have their mate with them. And so those kind of things, I think, are crucial for us as technologies to think about. We see a problem, not necessarily a problem that is going to make us a ton of money, but we see a problem that is really important for people's. psyche, for people's feelings, for loneliness. Who wants to be lonely? Nobody. And if we can't do this through pure social interaction, what's the next best thing? We have, or we, Amazon has Astro, but there's more robots like Astro. And they'll just run, they'll go search for you

through the house and ask you if you've taken your medicine. families or people that have these kind of robots, their medicine taking goes up 80%. Why? Because they're being continuously remembered. And it's, and actually that way predates this, is early on when we launched Alexa. I got a story from a man who was suffering. from early dementia and he knew what was coming.

Speaking of Alexa. That's her. And there she is. You can't help me was that. I know that. So the only thing he had to do is put a post-it note on the devices. It's called Alexa. Right.

Because one of the things that he had noticed. was that during the day his caregivers were getting more and more irritated with him. Why? Because he would ask him the same thing over and over and over again. What day is it today? What day is it today? Because he had completely forgotten. And the thing with her with that, let's say, voice assistant was that she could ask it the same thing over and over and over again.

And it would never get angry. it would always be happy and it would always give the same answer. So there are technologies that we can develop to help people have a good life. And I think if you can do that, spam filters, fine. Sure. Yeah, right. What I find interesting with this diet experiment is, and something I don't think I heard you say here or in your post, is we're actually, to some degree, programmed.

I mean, I know I had a lovy of sorts when I grew up as a child. I know that I've got two boys that are growing up, and they both have a thing that they were given at a very young age that is not technology, but it's about the best abstraction you can get. And a lot of people have had it in the last 50 years, even to 100 years, but even more, is this idea of a companion that you draw an affinity to. Now, this thing does not tell me about medicine, obviously.

It does not remind them about things, but it's this object that they attach to and cling to. So, I mean, how many kids don't you see walk around with a little rabbit after 10 years? Yeah. Yeah. And you mentioned the dinosaur. I mean, my kids would no way ever destroy their thing.

Like, it's their lovey. It's their, it's their prized, you know, friend companion that has, you know, no humanity whatsoever, no ability to speak, no true reciprocal affection even. But it's theirs and it has been theirs and it's their affection point that is uniquely theirs and only theirs. So even from a young age, they've had this companion. So it doesn't surprise me that your prediction is maybe that as an antidote to loneliness. Well, I don't think it's an enderous to loneliness. It is an insistent. I said antidote.

People are still lonely. Yeah. Yeah. Well, I have seen elderly ladies with a fake cat next to them, stroking the cat and the cat meows and the old elderly lady is extremely happy about that. Yeah. And so, why not? Yeah.

And the fact that we have had some, or a phase in our existence where we think like, oh, no, all relationships need to be human to human. Well, maybe not. Yeah. Maybe there is more to that, especially because, I know, we are getting older. I mean, look at all the baby boomers. Everybody over 65 or over 70. there's a very large group of people that is growing older and older and older and come with a new set of problems.

Can we solve those as technologies? Can we make a contribution to that? That would be great because I think building technology is cool but helping people is much better than that. Well, friends, I don't know about you, but something bothers me about GitHub Actions. I love the fact that it's there. I love the fact that it's so ubiquitous. I love the fact that agents that do my coding for me believe that my CI CD workflow begins with drafting Toml Files for GitHub Actions.

That's great. It's all great. Until, yes, until your builds start moving like molasses. Get Up Actions is slow. It's just the way it is, that's how it works. I'm sorry, but I'm not sorry because our friends at Namespace, they fix that. Yes, we use Namespace.com to do all of our builds so much faster. Namespace is like GitHub actions, but faster, I mean, like, way faster.

It caches everything smartly. It casts your dependencies, your Docker layers, your build artifacts, so your CI can run super fast. You get shorter feedback loops, have your developers because we love our time. time and you get fewer, I'll be back after this coffee and my build finishes. So that's not cool. The best part is it's drop in. It works right alongside your existing GitHub actions with almost zero config. It's a one line change.

So you can speed up your builds, you can delight your team, and you can finally stop pretending that build time is focus time. It's not. Learn more. Go to namespace. dot s o that's name space dot so just like it sounds like it said go there check them out we use them we love them and you should too namespace dot s oh i think we leave it right there and turn to adam's favorite topic quantum quantum quantum i'll say it three times the headline for this particular

prediction is quantum safe becomes the only safe now we've covered quantum computing here in the past and it's always been right around the corner it seems like it's always around the corner but your contention with some data to back it up is that that corner is getting closer and closer and it's maybe not five years away like it always has been or 18 months away it's actually there's things happening and it's like we're getting there we're getting there can you tell us more well i i do think more the most important thing is that i don't necessarily want to put a timeline on it i think in the sure two years ago three years ago i think i said five to ten i think five is now more realistic but i've seen a number of problems arising that i think uh we need to be aware of

and take action on before that i mean it's not about i mean we we run bracket which is together with Caltech is an institute about quantum and I see different institutions making massive progress in terms of error correction, networks, things like that, that actually make me believe that it's no longer, how shall I say, a vision. It's an execution now. Now it becomes an iterative process. in improving the technology that we have created step by step by step until it becomes workable. Now, there is a ton of other problems to solve. How do you program these beasts?

I mean, how do you do that? How do you do with something that is not necessarily 100% reliable? So how do you do that? So there's a ton of things that still has to be solved before we have production quantum machinery but one of the things that these things to end they won't solve necessarily new problems they won't so they'll solve existing problems much faster there are things that would now take a thousand years and compute will then take a year or six months or two months yeah that's the premise of quantum Well, the one thing that then will be able to do really, really, really, really fast is decrypt.

And so anything that is, let's say, is this elliptic or the usual RSA encrypted kind of things that are perfectly fine now. Because it will take way too long to decrypt. then will be decrypted with a snap of your fingers. So the reason for the prediction actually is to put in people's minds that we need to start taking action now if we want to be protected five years from now from, let's say, the ability to decrypt anything that we have now. Now, all the hypers, Google and Microsoft,

Amazon, whatever. We all have been building there's a post-quantum cryptography. We've been inserting that into our front ends. And we're doing all of that. Unfortunately, we're not the only players

in this world. There's lots of people with their own data centers still, with their colos, with their, whatever. That will have a hard time protecting themselves. That's one thing. So really getting it into people minds that they have to start taking action now if you

want to be safe a number of years from now when these things becomes reality. The other thing is that you can't just lean back right now because a lot of the breakings or data stealing or whatever you call it, is basically data harvesting. Nobody is interested, basically encrypted data is being retrieved and people don't try to decrypt it. They'll wait till the machinery is there to be able to decrypt. Now, well, your medical data now will be decryptable in a few years from now. That's not cool.

And so, and with that, I don't necessarily mean script kiddies or whatever. Yeah, it's really state actors. It might be, you know, commercial actors who are interested in that kind of data, even if it's five years old, you know, and I might have all the relevant data for them to take action on. That's the scary part, really. the famous TV show you may have heard of this Warner is called Silicon Valley

and in season six I'm going to spoil it for some folks just a little tiny bit let's just say and you may know this because you may be a fan or maybe you're not because it's too close to home who knows is that the AI they build

or accidentally build some degree they give it a task and it essentially goes out of their bounds that they plan for and breaks encryption. So this is predicted in a TV show that's now ended, but very much prophetic in a lot of its satire. And that's been my major concern with quantum, is that in the future, we have a machine that is, as you said, it's not deterministic or unpredictable in a lot of ways. It's also

unpredictably predictable in the fact that it's unpredictable. But it has so much possibility that that's when encryption will essentially go away, and the world we live in that is safe because of encryption is no longer safe. That's been my major thought around that, really, for quantum. There's, of course, different types of encryption. You have lattice-based encryption, and you have that. So there are types of encryption that you will be safe from sort of these type of decryption. But you need to start installing that now. You know, Lib SSS was 600,000 lines of code or something like that. I mean, you can imagine it because it had all sorts of different types of encryption in it.

At some moment, we started realizing at Amazon that running all of her contents with Lib SSS was a major vulnerability. And so we implemented TLS with a limited set of encryption technologies and just put that in front of S-Grieve, for example. First of all, much cheaper, but we made it open source and we made sure that it is fast and inspectable. And more important, written in a way that we can use, use automatic reasoning to see whether or not we protecting ourselves. So the most important

part, I mean, now you hear more, more about automatic reasoning with people like Byron Cook who have been working at Amazon for the past 15 years on these kind of problems, proving to ourselves that we are protecting our customers. That's not a press release that is going out or something like that. But there are technologies that, you know, you can use to sort of prove to yourself. As S&M, Signal to Noise, was the original name of the open source project,

was purely also intended to get post-quantum encryption in the hands of everyone. Now, I told you, now you got scared about the, And the data harvesting, let me scare you a little bit more, probably every of your home devices hasn't been updated for a number of years. Yeah, they were in Linux, whatever version. No automatic updates or anything like that. Your automatic garage door. Yeah, it might have a little bit of encryption in it between the key and you. or think about hotel rooms. Your key is an encryption encoder.

And so there are so many small parts. I mean, first of all, by coming back to these house devices, I recently looked at one of the house that we have, how many devices were on the network. And that was something like 56. There's your Sonos, and there's your garage, and there's your light opener, and there's the thing that keeps track of this, and all these kind of things. Are all these things up-to-date?

Are all these things protected? Are all these things up to new standards? We have a number of years to get this done, but we do need to start now. So the task today is what? What is the task we could do today? as developers, as maybe potentially inventors. What should we do today to plan for quantum? That really is the question in there.

First of all, and there's a number of areas. I mean, there's just using newer encryption technologies, but you can imagine that, you know, something, some piece of data that you encrypted 15 years ago, you're not going to go through all of your data and re-encrypt it again. But, you know, are there gateway possibilities that protect you from that? Are there other ways? Let's use our inventive brains and find solutions to this, because I believe we can,

but only if we make it a focus. And the focus is not that quantum is five years away. The most important part there is that we need to keep ourselves and our businesses safe. And if that happened to be quantum or if it happened to be an alien spaceship that lands from whoever and suddenly it turns out to be able to encrypt all our data, decrypt all our data, that doesn't really matter. We need to start protecting ourselves. How sure can we be that quantum systems?

safe technologies as we define them today actually will be quantum safe yeah well do we know do the work do the work i was hoping you already did it warner yeah yeah no no it's just do the work you know that no no that that is it the same goes for everything now let me take a whole the different example okay when at amazon s3 we went from eventual consistency to strong consistency. That was a major change that impacted every piece of technology. Now, how can you make sure that she touched every edge of this particular problem? Because it is either 100% or not.

So, automatic reasoning plays a role in that. We are annotating, annotations, TLA Plus, other technologies that you have to prove to yourself that what you've built actually does exactly what you wanted to do. So definitely, I think, algorithmic reasoning is continue to play a more and more role in giving us a secure feeling that we are on the right. right path. Yeah, the same is, for example, we recently launched Cairo. That is the spec driven IDE for, with an AI system.

Yeah, we did a show on it. Yeah. We use automatic reasoning there to limit the number of hallucinations that can happen. Yeah. Not that automatic reasoning knows that terribly much more, but it knows A square plus B square, C square, is probably a triangle. And as such, it knows all the triangles in the world. If then this LLM says, this is a triangle,

out of the magnetic reasoning, you'll say, no, that is not a triangle. And that is just a simple example, but we start building more and more technologies to prove to ourselves that we are doing the right thing. Well, let me just take a quick moment to say, we talked Wikipedia last week, Orner, and I called it perhaps the eighth wonder of the world. And if it is the eighth, I would say Amazon S3 is probably the ninth wonder of the world. So just a quick moment of props to you.

And thank you for what is an amazing, amazing piece of technology that's unlocked so much for so many businesses and people over the years. S3 is wild. Yeah, yeah, I am. You know what my first programming languages were at school? No. Cobol, 68,000 assembler, and Pascal. Okay.

No one of these programming languages, anybody writes anything anymore. Although maybe I could make money as a COBOL program. I'm not too sure I'm doing well in 68,000 anymore, but things change over time. and we as humans adapt. We're often ahead of the game. You know, we went from things like Pascal, which probably was sort of the first version of a little bit of structured programming

to C++ plus and Java and Python and just the whole world. None of us panicked there. We just learned a new program. language. It's not like it's not like that we're suddenly

frozen in our chair and now we have AI AI tools to help us building the systems that we do why should we start to be scared suddenly? Why should we suddenly become we just learn

to use new tools? Because remember it's you that built, not the tools. It's also you that own it, not the tools. If the tools is built something that, and you're in a financial company that suddenly no longer makes you compliant,

it's pretty hard to say to the regulator, oh, that was the AI, that was the tool. Yeah. No, no, no, no, no. It's still your responsibility. And as such, I think that sort of we, no matter how good tools we build, we still need to educate ourselves. We need to continue to keep ourselves on a path of education, learning, continuing to learn. And looking at you too, you're not the youngest. So I'm pretty sure that before you wrote pie with, something in Python, you wrote something in C++, or in ABC. Yeah? Oh, yeah. Well, I wasn't learning Python or going to the next programming language, my favorite being rushed at the moment.

You know, but there will be something else after Rust. Sure. We won't panic suddenly and thinking that the world will collapse and all our jobs are going to go away. No. No, we make sure that we educate ourselves. And crucial in all of that is curiosity. Yeah, if you just want to lean back.

So if you just want to lean back and just, you know, write some code and things like that and don't want to learn the next language or whatever, then probably you've always been in so long. That has nothing to do with AI. You have to, because curiosity leads to learning. And these, those two things are, remain curious. Yeah. What was it? Have you guys seen Ted Lassau?

Yeah. Oh, yeah. The TV series. There is this brilliant dart scene where he froze 170. And he tells this story that I saw this quote by Walt Whitman. It was painted on the wall there. It said, be curious, not judgmental. I like that. And he said, I suddenly realized that nobody had been curious in who I was. They were just bullying me just because they could. And curiosity is such an important part. Being curious about AI, being curious. Being curious.

about quantum. Be curious about how defense technology actually transforms into civil technology. If you're curious, you'll find out. Yeah, and I think if you look at young kids, they're all massively curious. Yeah? Until we send them to school. Yeah? And then we make them conform to one standard. Yeah, they all need to become the same. And regardless what their mentality is, regardless what their level of interest is, I think everybody has the ability

how shall I say, to become special. Well, there is a, I think one of the talents, and given that what we have, we have the Renaissance developer as one of the points in there, in the Renaissance, the core of Rememasons humanism was that the belief that humans are limitless in the capability of invention. As long as you keep their curiosity alive, the moment they're no longer curious, yeah, they'll lean back and it's over. And so I believe that that's really, it's a, they're limited in the capability, we are limited in the capabilities of inventing, of new, do new things. If we're just being allowed to. Yeah, I don't disagree. you're hit on a couple of your points the renaissance developer and the personalized learning which i think is hugely uh opportunistic or a huge opportunity for so many because like you said we

sometimes by necessity we push all of our children into a square hole you know regardless of their educational shape and they must come out conformed to whatever standard and i do think that technology is allowing us to break out of that conformity to a certain extent provide personalized, customized education leaning into what the kids interested in

and playing to their strengths. Yeah, but also, even also just learning maps. I remember I have two daughters. They're now in close to the 40s. Don't tell them that. they got to know that right one of the things so i'm not not american i'm not the u.s i come from the netherlands where high school has 10 different forms depending on sort of what you're

you go to a school to become a welder or you go to a school way go to university and there's a whole range in between. And then they come to the rest and the kids have to go to high school and all kids have to go to the same high school. They have to take the same classes. They have to score good in the same classes. And when one of my daughters in the first time that she got into MAF cars, there were four guys in the back of the room that were four years older than that she was, still had failed maths every time, but had to still sit in that class. Disruptor fan, well, you know the stories. Yeah.

And as such, I've always been a real fan of Ken Robinson, professor from the, from Oxytok or Kinwet, whatever, whatever, from the UK that moved to Los Angeles at some moment. And I've been a real fan of him because he really looked at that instead of that we let our kids bloom and explore and become really what they can become, we turn them into factory workers, all doing the same thing. And that kills all the possibility. in what these kids have in them. What do you see coming out or happening in the education space right now that leads you towards this prediction? Because you're predicting, you know, infinite person personalization

and these things are there solutions out there today or are their companies working on it? Well, I think there's enough companies working on them. But what I see more and more is that these kids are doing it themselves. Yeah, Gen Alpha, the current kids that are in high school or too high going to high school, they really know how to use AI or AI assistants. And they know how to make a curriculum for themselves. And they make, they make, they know how to make something from which they can learn.

Yeah. That is really a tool for them to build something that they just want to do. It helps them build things without having to have to go through a complete textbook first. Now, I still believe that you need to be able to do maths before you should be allowed to use a calculator. Yeah, that seems like a reasonable kind of thing. But providing, if you're really interested in, let's say, how in the 1600 in Italy, from small states this became one country

for some reason I don't know kids have immense curiosity in anything you can build yourself a curriculum that you become not like you become an expert

but that you know everything about it and this is something you have suddenly you have control over that now there's always two sides to this the other side to this is that, and this is actually something from Ken Robinson again, he says, there are no good schools, there are only good teachers. And he says, and what do most teachers

spend their time on, grading homework, doing administration, all these kind of things that actually have nothing to do with education. So, yeah, that, that seems to me like a standard technology solution problem. Yeah. And whether you solve that with AI or whether you solve that with some, just some simple sequel, fine. Yeah, that doesn't really matter. I believe that there should be a lot more support for teachers,

such that they can focus on things that is important. Individual interaction between a teacher and a student has so much more impact than the teacher just standing in front of the class. Yeah, by the way. Ted tells this wonderful story that he's a teacher, and he tells his story that he's giving drawing lesson. And there's always this one girl in the back of the class that is never interested in anything.

But now the drawing lessons, and she's fanatic. So he goes to the back of the class and looks at what she's doing. And he says, what's she doing? And she says, I'm drawing God. And he says, but nobody knows how God looks like. And she looks at him. He says, but in a minute, they will. That's funny.

Yeah. Yeah. I mean, do we think, here's a blank paper. Why don't you draw God? Yeah. Yeah? No. Yeah.

But this is a kid. There are no boundaries to your curiosity and to what you can. All these things that these kids feel as constrained, are put upon them. Yeah, it's a double-edged sword with this curiosity point with children. I think, especially with AI, unfettered curiosity at a young age with such a powerful thing is not that smart, but I've personally experienced my ability to be curious and to explore at my older age.

And I like to translate some of that to the younger generation, because I agree. I think there's this idea of like just in-time learning when you learn about a new thing and there's curiosity there to like accomplish a goal, but you can't really accomplish that goal until you've gone a little deeper here, here and here for context. I think that's kind of the new term or the new word of maybe potentially 2025 is context.

Sure, AI and maybe even agentic, but I think a subversion of that is context. Yeah, yeah, yeah. You know, this just in time learning is really, really fascinating. And you must have been to conference or whatever, through some tech conference at some moment where there is a, someone describes something about a new cache algorithm. Yeah.

And they look at that and I think that shouldn't work. Yeah, or actually an Amazon has something called the Builder's Library that's available for everyone to look at. These are articles written by Amazon most senior engineers. And there is this one article by Colm. I couldn't get my head around because I've grown up in a time that you minimize the number of bites on the wire. One more bite means more time, more delay, whatever. That's no longer the case.

You know, he describes this system in which if the receiver always has to do exactly the same number of things, because there are ten things in this packet, but eight of them may be empty. But everything is just standardized, and you don't need to think about it, and the resilience grows up. And so it took me a long time to get my head around that. But the first time I read the intro to the story, it immediately picked my curiosity, and it took me a long time to learn what he really did. really try to achieve there. And so I think, you know, no matter how old we are,

there are moments every time when you get sort of, oh, how would that be? I live here in a place where there were a lot of electric scooters. Yeah, the ones that, that, was it, Lyme that Uber bought and some others and, and whatever. And I've always seen them, but I always was interested in, curious and how would it be to where I want? Well, now it's just a click on your phone and you jump on it and you go to the supermarket to buy your groceries and then you come back. you know and it's just fun and so but you know curiosity is really one of the most amazing things that is uplifting in our life so there's only one of you and there's only so many hours in the day and really only so much focus to go around so my question to you is what if you had a teammate who could work just like you with all the context you have but a little faster

that's what notions new AI agent feels like notion already brings all your notes, all your docs, all your projects, into one single connected space that just works, seamless, flexible, powerful, and actually kind of fun to use. But with Notion Agent, your AI doesn't just help you work, it finishes it. Here's what makes Notion agent different. It can do anything you can do in Notion. It taps into your workspace, the web,

and the connected tools like Slack and Google Drive to complete assigned actions end-to-end with a single prompt. It forms a plan, executes it, and will even reassess and try again. if it hits a nag. Now, creating new pages, building databases from scratch, summarizing entire projects, it does it too easy. And since it's all inside Notion, you're always in control. You tell your agent how to behave. It remembers and updates automatically. Now, over 50% of Fortune 500 companies use Notion and fast-moving teams like

OpenEye, Ramp, and Versal, they're using Notion too. And we use Notion. Send less emails, cancel more meetings, stay ahead. It's too easy. So if you're ready to try, your new AI teammate, Notion agent, head to Notion.com slash changelog. Get started there. That's all overcase, notion.com slash changelog. And of course, when you use our link, you're supporting our show. And we love that. And by our friends over at Nordlear, toggle ready, network security for modern teams. Yes, if you're running a team, whether it's a five-person company or a 500-person company, network security should not require a dedicated IT department and months of configuration. Nord layer is a toggle-ready network security platform that combines VPN,

access control, and threat protection into one easy-use platform. No hardware, no complex setup, just secure connection, and full control in less than 10 minutes. The cool part is it's built on the standards of NordVPN, and it's powered by their Nord Links protocol based on Wiregard. NordLair is designed for businesses, not consumers. It's based on zero-trust architecture. Only the right people have access to the right resources under the right conditions. You get granular control over who access is what, from where, and on which device. And the chef's kiss, the cherry on top, so to speak, it works on all platforms, Windows, MacOS, iOS, Android, and of course, my favorite Linux. So the next step is to go to Nordlayer.com slash the changelog, yes, Nordlayer.com slash the changelog.

Use our code, changelog hyphen 28. Yes, the hyphen's important. Changelog, hyphen, and 28. Guess what? 20% off Nordlair yearly plans using the code change log hyphen 28. It's valid until December 10th, 2025. Once again, Nordlayer.com slash the change law. It makes me think about this next notion, which I'm not sure I agree with, but it's been potentially the phrase or at least the sentiment of the last year or so, mainly. around fear, uncertainty, and doubt, which is developers are dead, that we're a dying breed, AI is taking our jobs, it's not, it's going to replace us, and you seem to push back on that idea through being a renaissance developer.

I do think if your job was only being, and that sounds a bit negative, being a code monkey, yeah, just tap it. All you do is code. That's all you do. You don't think, you don't play right. Yeah. If that would be, but I think most of us, I've had to think about some of the stuff that we built.

We had to think about algorithms. We had to think about tradeals. You had to think about sort of, you went over to your colleague on the other side, and you said like, you know, this is what I need to solve here. But I think there's a cash overrun, and that will brown out,

then do you think that? So shall we set up a test environment and see whether that works or not? Or, you know, there are so many parts to developers, development that will require our brains. And that's why I picked the word Renaissance is because before the Renaissance, there were the Dark Ages. Yeah?

1,000 years, basically, the black death, the plague, all these kind of things, and everything was ordered by basically religious institutions. And that changed at some moment where people became interesting again in all the amazing things that have been done a thousand years before by the Arabs, by the Romans, by by the Greek, and so where they suddenly completely started to change. And there is this word that is sort of associated with the Renaissance, that is Polly Mouth. And many people think that that's people, that is someone who can do mathematics in many different forms.

No, no, no, no. Poly means Moth. And Matinealia means learning. much to learn. And if you look at the people in those days, definitely, they suddenly started to become interested in many more things than just only the one thing that they were specialized in. Da Vinci was not just a sculptor.

I mean, he built airplanes, well, they didn't fly. But he did build them, yeah, and just trying. I mean, think about all the people in those states that suddenly allowed their curiosity to come out. Now, and I think as developers, there are a number of things that we will need to we encounter. First of all, I do think we've been learning for 50 years now, or whatever, how long we've been doing computer science, probably 1960s so yeah 60 years yeah so we've been learning programming languages for 60 years we've been going from what was it from pieces from main for main frames to minis to PCs to cloud to whatever everything required a new approach to building or enable this to do new

things, you know, before, before Cloud, you may have only had one Kolo and never would be able to build a replicated database, but then CloudGamer suddenly you could replicate your database. You need to learn that, how to do that, but still, it is. And definitely, I believe that it's not just sufficient to, and now I need to be careful because I don't want to give away too much of my keynote. Yeah, go ahead. Spill the beans. I'm just kidding.

I believe to be successful, not only in the future, but also now, is to not only have a deep specialism, but start to become interested in the broadness as well. It's called a T developer. Yeah. It is someone that is not just a database expert. Not just a database expert, but he knows about the other types of application. He knows about all the programming languages. He may even be able to help his colleagues on the other side.

Or at least can have a conversation about it. Understands the business principles of the business that he's actually willing to. another area that I think new that these type of community the newer type of developers the the Renaissance developer needs to be an excellent communicator and with that for example I mean if you look at the Amazon website just imagine that for you there are three there are some parts of the Amazon website that always need to work because otherwise we can't sell anything browse, search, shopping cart, checkout, and reviews. Because if there are now reviews, people don't buy things.

Five things that absolutely need to work. And then we have all the other things that actually are really important as well. Then we have recommendations, personalization, whatever. And then there is stuff that is just nice to have. Now, take, for example, best sellers. list. Now, as a technologist, you need to be able to have a conversation with the business and say, how many nines do you want behind that? Oh, you want four nines behind your best seller list? Well, that's going to cost you so much. And then it becomes a business

decision how I am going to implement this. But I need to be able to explain to the business, what the risks are, what the costs are, how resilient can I make this. But without that kind of communication, you're not building the things that you want for your customers. So a very important part of the Renaissance developer is to be able to become to communicate in exactly the right way not not only to the business yeah but also with your customers you know what are you building how does this interact with this you know if i show you a picture of the thousands of microservices that builds out amazon dot com yeah if i do something here yeah what's happening there

is there a relationship between the two? So system thinking becomes crucial as well, not just about your little piece that you're building, but what impact does that have on other things? Yeah, or, you know, I'm happy going away, my database is running, my code is running, everything is happening, transactions are playing through. And suddenly the number of transactions doubles.

But there isn't a doubling of the number of the customer on the website. Who the hell is doing this to me? And as such, being a specialist in one particular area and just staying there is safe. Yeah, it's fine. But you live in a bigger system. You know, if we think about, and that is, that really comes out of 60s, 70s, Medellum Meadows began what's something called system thinking. It is this, in nature, you know, if you remove one piece out of that equation, of the complete nature

equation, you ruin the overall thing. At some moment, how was it, a yellow stone park. There were too many wolves. They removed the wolves from the park. Oh, they were killing all the elks. So they remove all the wolves. And then the number of elk grows like mad. Backfire.

Yeah, which starts eating all sorts of parts of yellowstone. And so not until they, we introduce the rules, the balance gets back. Yeah? So on her thinking, on systems thinking, is something that we as the future Renaissance engineers need to be part of as well. We need to understand that the one thing that we're working on at this moment has impact on all the other pieces. around it. We need to be able to communicate with the others. We need to be able to see how we are part of this bigger problem. And you know what? And it will be fun. It will be fun. I absolutely think. And so one of the things I think that I won't say, hey, let's say,

one of the most less interesting parts I always, in the beginning as my life as developer thought, were code reviews. It's a bit like standing in front of the class. The whole class get the chance to say like, you're an idiot. Right. No, but code reviews are fun. You know, sometimes you think,

there's mistakes there, or sometimes. And you go like, wow, that's an elegant solution. I wish it thought about that. And code reviews are super important. for junior engineers. Because the senior engineers are in the room and discussing the code on the screen there and you're learning on the spot

from these senior engineer, from all the years of experience where someone will say, yeah, I tried that when we were building product X, Y or Z. And it didn't work then, or the problem at that time was that. Now, where did the code? on the screen is generated by you or by some AI agent, doesn't really matter, we still need to do

code reviews, yeah, because we have ownership. And whether we used a tool to create the code that we're looking at, or whether we actually wrote it ourselves, we're still responsible for we still own it. And so those things don't go away. And as such, if your system has been created by one of the AI tools, and as I said earlier, you know, you are responsible for it, especially if you're in life sciences, if you're in healthcare, if you're in financial services, in whatever,

you cannot only rely on the fact that oh but AI build it you are still responsible for what you deliver and that means you need to be able to look at it and a search generation of code may go faster refuse will probably go slower because this is not code that I've written now if it's my code I can go stand up in front of the class I say oh and this is why I did this and this is why I did that. But if a complete unknown, unhuman entity created this code for you, it takes longer to get used to it. And so, that's not bad.

I mean, none of this. Don't take this as that I'm sort of cracking on AI as being. No, there's this parts that will never leave us. we need to learn the next steps we need we continue to keep ownership of what we do we continue to need to learn

the next language the next tool the next system that we built and you know we need to know more than just that one piece we need to be more than that and my favorite

and he'll he will he will he comes back in all every presentation that I gave was Jim Gray. He won a Touring Award. This is the guy that actually built System R at IBM, the inventor of transactions. Jim was brilliant. Absolutely brilliant.

There's an article, if you want to look it up. It's called 20 questions to Jim Gray. Jim said, give me 20 business questions or questions that you as a user of a database would want to answer the database to have the database answer, I will build a database for you. And by the way, one of the most interesting questions is he walks into the room where the big machinery is and he hears the disks rattling and he looks at, he looks sideways and he listens to this, he says your database layout is wrong. That was something he could hear.

Now, he was not just a database expert. He had a lot of other skills in other areas. And so he was what we would call a T developer. An I developer is someone that knows only one thing. A T developer has, you know, number of know know know, number of knowledgees in other areas of whether it's computer science, whether it's from the world, or whatever, it's a broader person than just only the database developer.

And I think in the future, this is something that we will continue to harness. You need to know more, you need to be able, because how can you communicate with your colleagues that build other parts of the system if you actually don't know anything about what and how they're doing it. This idea of co-review, one thing you're thinking about, you're making me think about is, at least the sentiment it seems today is getting through code review. It seems like you're suggesting, even though it's painful, not so much pausing, but not

getting through it just to deliver the code to production and correctness, but to understand why the code is what it is, and especially if it's AI generated and the human in the loop is less informed specifically about the code itself, that it's more of an artifact of a plan, not an artifact of actual implementation themselves. Yeah, and I'm getting old. I just had a great answer to what you wanted to do. And I completely lost it.

I do think that we've changed over time as developers. I mean, I've written VB. No, I don't think there's anybody that writes Visual Basic anymore, is there? Not on purpose. Yeah. Fix a bug maybe in Legacy Code. Yeah, something like that. But, you know, but it was fun when we were doing it.

And, you know, I wasn't thinking about five databases replicated over five different data centers throughout the world at that particular moment. I was trying to fix my Excel code. And so we've all grown as developers over the past 10, 15, 20 years with all sorts of new tools. We have a new tool. There is a particular point, a particular small problem I have with it, is that if you look at all the other tools that we've developed, we took some time to introduce them. We had learned people, we told people why we were building the tools, how we were building the tools, how they could best use them. Yeah, the typical early adopters, and you had to cause the chasm,

and then you had the early majority and things like that. And it would take time. But you would learn, you would teach your users why and how that tool was there. With some of the current, the new generative AI tool, they were just dumped in people's lap without telling them what it could do or what it couldn't do. And, you know,

there isn't almost a CIO that I meet today when I'm traveling that asks me, he says, what should I be doing with AI? And I go like, well, my excuses is very inappropriate to answer a question with a question but why are you asking me this and just yeah but those guys next door you know they will be ahead of us or and I say are you really certain of that and then maybe because you're a little bit older you know you start to

drive down into what is actually the problem the train to solve with this technology and is this the right technology for for this every week we see five new models or 10 new models suddenly we went from regular LLMs to reasoning LLMs I think as a business there is no shame in putting, hitting the pause button for a moment and say, why don't we get ourselves educated about all of this? Yeah, and not just us as technologists, but also the business. Yeah? Because at this moment, quite a few of our architectures are being determined by the media, not by us. Yeah? That shouldn't be the case. Right. we, together with our business, should determine how our architectures should look like,

not because the newest thing. And of course, and that's the task of the media. And the task of the media is also to write negative headlines, even things are looking up. That company is massively behind that company. Yeah, you know what? AWS was never a consumer company. So we don't build consumer products.

We build tools such that you can build your chatbots. Right. If Amazon, if AWS doesn't have a chatbot, means that we're out of business in AI. It just means it's not our business. And I think sort of that's really crucial in all of this, that we take time to learn the technologies, the capabilities of the technology.

and where it can help us in our businesses. And, you know, if we also give it to a whole bunch of Gen Alpha kids that can go do wild with it and build our own educational curriculum with them, great, let him have it. Yeah? It's a bit like us 50 years ago with hammer and nails and building your own thing up in the tree, you know, and now kids do different things.

But, you know, we need to give them the tools for it. And, but definitely in business, I think, there is no shame in hitting the pause button and spend some time on education and making sure that you and your engineers are making the right decisions with the right tools. There is no hurry. There is no company will go out business in a company two months because they weren't using AI.

Well, you heard it here first because very few people are saying that out loud. And while I agree with you, and I think that it is a problem. I wonder if we have an appetite as an industry to address said problem. I think we're just barreling forward and we'll see what happened. But I also agree when you say give the kids the tools. Give the kids. Oh, it's a good one. one of the other things

I didn't make it this time into the predictions. Did they do it last year? One of the things that I think last year is one of the things I talked about. One of the things that worries me is that

parents driving in their car, the kids are in the backseat they're getting that and are you going home, when are getting home and that was how we used to drive home. Now the kids can iPad and become YouTube experts and they're quiet in the back. And the parents love it.

But she set kids up from four or five years old to get dopamine reactions. They know how to manipulate YouTube continuously. And it's not only that they are saying they're getting this dopamine high. There's something else called dopamine too that actually removes your willpower in these kind of cases. Whenever you stand at the best bus stop next time, look at how many people pull their cell phone out,

even if they know that the bus is three minutes away. Nobody wants to be bored anymore. Nobody wants to just stare out in the distance and use their brains. you know and we can't help ourselves because because we've we've ruined ourselves and if we do this with our young kids we will have an epidemic on our hands 10 15 years from now in terms of addiction in reason sorry guys I'm I keep yapping no we appreciate it we know you got to go didn't quite make it to your fifth but uh great great predictions thank you for sharing your

extended thoughts on them and giving us some time today okay well i i really enjoyed it guys come on watch my uh keynote the last keynote i reinvent i'm doing the street 30 on friday and first day nice after me everybody can get do whatever they want you're the closer you're the closer. I am the closer. Awesome, Werner. Well, you're welcome back anytime. Coming close.

Yes. Okay. Bye-bye. Thank you, Warren. There you have it. Werner Vogel's predictions for 2026 and beyond. He had another one about defense technology that we didn't quite get to find the link to his article in the show notes. So, do you agree with him?

Do you disagree? What do you think will happen next? Let us know in the comments by joining. our free community and chatting with us in Zulip. That link is also in your show notes. Thanks again to our partners at fly.io and to our beatfreak in residence. Breakmaster cylinder, who's hard at work remixing your state of the log voicemails. We have a couple trickling in, but we need more. Send us one. Why don't you? At changelog.fm slash sotl. That link is also, you guessed it, in your show notes. All right, that's it. Goodbye for now. But we'll be back on

Friday with Nick Neesey talking Anthropics acquisition of Bun, source graph and amps splitting into Nick's latest Claude code shenanigans, and what do we make of the new browser war? It's a good one, we'll talk to you then. Game on