Dan Bricklin: Lessons from Building the First Killer App | Learning from Machine Learning #14

When VisiCalc first came out, you could explain it or show it to people, okay? And if you showed it to a computer person, they look at it and say, what's so special here? I can write a program that does that and I can do better and stuff like that. You show it to a regular person, a normal person, they say, computers can do anything.

They can forecast the weather. What's so special about that? But if you showed it to people who actually did that stuff, They would start shaking and say, here's my credit card. Please take it.

Give me that. Give me that. How did the best machine learning practitioners get involved in the field? What challenges have they faced?

What has helped them flourish? Let's ask them. Welcome to Learning from Machine Learning. I'm your host, Seth Levine.

Hello and welcome to Learning from Machine Learning. On this episode, we have a very special guest. Dan Bricklin, the co -creator of VisiCalc, the first electronic spreadsheet and really the first killer app that helped launch the PC revolution. Dan has lived through every major computing platform shift from mainframes to today's AI.

Few people have a better perspective on what separates transformative technology from hype. Dan, welcome to the show. I'm glad to be here. Thanks a lot, sir.

It's such a pleasure to have you. Let's get right into it. So what initially attracted you to computer science, computers in general? Well, as a child, this is back in the 50s and 60s.

I was, you know, I love technology and, you know, mechanical stuff and electrical, electronic stuff and whatever. And I would read Popular Electronics magazine. That was my one of my favorites. So a computer seemed like I didn't know what it was really, you know, how they worked or how to make them.

But I knew that pretty cool stuff. And I knew some of the electronics of how to do whatever. But so I was interested in that. The first time I actually saw a computer is I was quite young.

I was in early and grade school and I visited my grandfather, editor of a small newspaper, and I visited him at work and they had a punch card machine. IBM punch card machine they use it I guess for circulation for subscriptions and stuff and I was really taken by that how it could sort do the sorting and I actually built for a science fair project I built a rudimentary punch card sorter that you could Crank it and you know as the thing went through it It was able to tell and the magnet would open lift something that would go here or there depending on it So that that sort of got me wanting to get into it Um, and so that attracted me.

Uh, I was always, uh, viewed as a techie into space and things like that. Very cool. So then, yeah, we'll fast forward. You will see you did, uh, you started at MIT and well, yeah, but before, but I got into computers early.

I mean, I got into computers, uh, when I was like, uh, what, 15. My cousin came home from his school and they had just gotten access to a time sharing system. They were one of the few schools in the whole Philadelphia area who were able to do a version of Fortran called QuickTran using a remote terminal, type terminal. And I got to go over and learn.

He brought the manual home and I studied the manual and learned Fortran, enough Fortran too, and started programming. Like one of the first programs. that I wrote was we had learned in school patterns of English. We learned about noun, verb, whatever and that a sentence had this.

So I put in all these types of sentences and I put in a list of nouns and verbs and it would make up... It would make up sentences and stuff that were technically correct. I thought that was pretty cool. That's very cool.

So that's one of the first things. And then I built a graphing program. I think the thing that did graphing, you know, on the printer, the typewriter output. Then I would beg, steal or borrow computer time to be able to improve my stuff wherever I could find access to computing.

In those days, it was very hard for like a high school or anybody to get access to computers. But I learned how to be able to find where... There were ones that I could get access to. They're usually, you know, like the Bureau of Public Education in Philadelphia.

Right. I was like the only one who would go convince them to let me go down and use their 1401 computers. Yeah. I mean, it's so cool because it was like right as it was really becoming a thing.

So not very many people had access to it. So you were. Well, yeah, as these are all mainframes. Right.

I mean, I in late. Let's see. In later high school, I took a course, a summer course at the University of Pennsylvania, where the National Science Foundation had a programming computer course. And they had this 1130 computer that was brand new.

And I was able to program that and learn computer stuff at that course at the University of Pennsylvania, even though I was a high school student. And then I got a job at the person who ran that course. worked at the Wharton School at the University of Pennsylvania and worked in computational services. And I got an afternoon job in high school to be able to work there.

And I had access, suddenly I had access to computers through the University of Pennsylvania. Very cool. As a high school kid. Nice.

Yeah. So I was at the engineering school at University of Pennsylvania. So every day when I would go through one of the halls, I'd pass the ENIAC machine. Is that, is that?

Yeah. So I guess, yeah. Well, whatever. It was down the hall from where I took my course.

I knew that there was nothing to look at or anything. Yeah. Now it's just there behind glass and you can kind of look at it. So they built a shrine.

So they didn't have a shrine. They didn't have the shrine. I don't remember there being a shrine. I didn't see it.

I just knew that it was somewhere. I was at the Moore School, which is where it was done. Yeah, that's where I took a lot of my classes. It's kind of cool to be connected to the past.

I like looking at the history of stuff. So it's nice to be connected to the past. Yeah, definitely. Tell us a little bit like a mainframe computer.

Like, so does it have all of the same things? You know, how different? Those days? Well, in those days, I mean, they're very, very powerful for those days.

Today they're, you know, my watch is more powerful. Right. Microwave oven is probably as powerful. They were in these big, you know, the cabinets that look like big refrigerators and multiple of them, and each one had different parts in that.

Uh, now we might have, you know, half a chip or a fraction of a chip to be that whole thing. So there was the CPU, maybe a couple of them. There were tape drives, you know, that, uh, were used for storage for longer term storage. And there were disk drives, you know, and stuff like that.

And it produced a lot of heat and they had lots of big cables for lots of power and stuff. So the floors were raised. They had, so you could get, you could pull a, a tile off and get underneath and. you know put the wires down there between them so you wouldn't trip but also they could take air conditioning through it to get the heat off and the air conditioners are running you know you hear the noise when you go in the there was a computer room and you didn't go in there because you don't want to get any dust in there right or anything like that that might mess up the card readers and stuff so you know you did things outside and Plus, you don't want to give normal people, regular people access to that room.

They might mess something up. So that was what mainframes were like. They had CPUs, they had short -term storage and long -term storage and non -volatile storage and they did all the same. They had computer languages and you could connect terminals to them of various sorts to do visual stuff or printed output with paper printers that could print 10 pages a minute or something like that.

You know, and that's how checks were printed and and reports and stuff So it's a lot of what we have today, but just it was bigger Just if you take today literally within the last few days for most for any normal person getting it You have the new Apple has a new phone where all the electronics are basically are where the camera is a little strip where the camera is is the whole thing of the phone because they've managed to rise the whole phone down to the cameras and everything fit in a little strip and the rest is battery and screen but the basic idea of storage computation you know uh and display and input in and those are all different parts software that runs on them it's a stored program computer that that is you know, from an architectural viewpoint, it's very similar.

Right. It's pretty amazing to have seen the entire evolution and being one of the people that was accessing those mainframes to now see how that has all evolved into today. So some of the work that you did helped propel the PC, you know, the personal computer revolution. You worked on a word processor pretty early on, right?

Yes. Yeah, very early on, I was at Digital Equipment Corporation after college. At the end of college, I worked on a system that used a language called APL that regular people, but mainly engineers and people like that, could type in a simple program and get the results out of it. But then when I went to Digital Equipment Corporation, which was a large mini computer manufacturer, I worked in computerized typesetting.

and worked with editing for editing the articles that went through for connecting to wire service stuff. That's data that's coming in from outside. And then I ended up in their first word processor. It was a computerized word processor, a document oriented one.

In those days, there were two types of electronic with screen based word processing. There were just a few companies that made any. of those and one of them was was page oriented kind of like page maker or um powerpoint or something you laid out one page and then you laid out the next page and you position things within that as you type document oriented is what we're is more like word is what we're familiar with today. Ours used just one long thing of text and it broke it up into pages and cared about layout horizontally with tabs and decimal tabs that line up just right and how do you do that and I worked on that which was to be used by typists.

Then hopefully by managers and executives, but they didn't type in those days, but they would give things off to typists So that's I worked on that that software ended up on all sorts of different hardware It started out with a computer in the deck inside of a desk. The computer was a big and the screen Part of the desk eventually they sold it as a one -piece unit for about ten years as the deck mate Word processor that started as a deck WPS eight system. So and it looked like this.

Here is an ad for it. This is a typewriter. This was our system. That was on top of the desk, you know, and but you can see the desk with the and the floppy disks, the eight inch floppy disks that went in it.

So and it could do output like this. This is the functional specification for it. for this thing written in itself, in an early version of itself. So I got to, that was word wrapping, being able, and I wrote software that did some of that stuff.

There was a team of about four of us doing the software for that word processor. Before that, I had worked on an editing system for newspapers. I was familiar with some newspaper systems that were specifically for editing. You know two -dimensional layouts for giving the commands to a type setting system for doing the ads in a newspaper I got to see those that people had there was an interactive system called the Harris 2200 that you can't find too much about but was a groundbreaking machine that was basically page maker and hardware and That I was quite taken by and that influenced visit calc in a way Yeah, so getting into VisiCalc.

So, you know, VisiCalc, I mean, widely regarded as fueling, you know, the rapid growth of the personal computing industry. You know, Steve Jobs went as far as to say that VisiCalc is what propelled Apple II to the success it achieved. He says that in an interview in like the 1990s. Because it was only available for a year on the Apple II.

Because software takes a long time. Yeah, where do you want to start with visit calc? It's it's referenced everywhere right like in prediction machines The book it talks about like that's like one of the first steps towards realizing the power of automation You know change the way that it was interactive computing people were not used to interacting with a computer other than to put their card in and get money out of a machine, right? That was the main way that people interacted with a computer.

Here was a thing, a word processor, you know, maybe they had tried a word processor, but here is a thing that, you know, did things that they would have to do by hand they were supposed to, but it could do it so much faster and open up all sorts of worlds. So that That really affected things quite a bit. It let people say, oh, this is what an interactive computer is. It's not just games.

They did have access. They were playing Pong and stuff like that, Atari stuff. And so, oh, it can be useful in things that are better than anything else. Even a mainframe doesn't do anything.

This interactive, me give and taking with it and getting the output that I needed and it's custom to what I need. that just like a word processor. There were programs before word processors that would construct a will, would ask you questions and then write a will. But a word processor lets you write anything.

You know, that's a general purpose tool. Well, a spreadsheet was a general purpose tool for numbers, like a word processor was a general purpose tool for paragraphs or prose. For numbers and for also text, people often used it only for text to keep lists. And since you could lay things out any way you wanted, it was not strict columns and rows that all the rows are the same, and all the columns have the same one.

No, it wasn't like that. You could put anything anywhere, but you could take advantage of it. You could very quickly create things because it had just the right tools to let you do things very fast, because it was optimized for keystroke, minimum keystrokes, because I came from type setting where you're paid by the keystroke so I learned how to optimize for quick and I was competing against paper with a calculator and a pencil and paper and it had to be faster the first time you used it because otherwise you wouldn't use the recalculation if it was quicker you say oh I only need this once I'll do it by hand right so so that I think that said people wow Computers can be useful in so many ways.

What other things can they do? This is interactivity. What you see is what you get. The thing is you could build something and build on it and build on it and give it to somebody else.

And then all sorts of other software start coming out. Yeah. So you talk about it... You have like a some sort of formula about figuring out if something can be like a killer app.

You were kind of referencing it before, like, so you're comparing, like, what was it before, right? Like people were trying to do spreadsheets existed, right? It just was the idea that you could use a spreadsheet on, you know, on your machine, you could change one number, and then that would lead to cascading of everything changing, you could put in custom formulas, you could do all this stuff. I guess, as you were making it, did you know, like, did you know the magnitude of what you were creating?

Well, of course it should be. Whenever you do something, it's going to be great. You know, it usually isn't. Yes.

I knew it would be very useful, but people were not adopting computers at the rate we expected. Word processing was not taking off at that point. Right. Back in 1979, it was not going like that.

Normal people did not get it. It wasn't until the inexpensive. computers like the Osborne and the Kaypro, you know, that a journalist could buy their own computer to be able to take home and be able with a word processor at all and write things. And suddenly people could do their novels on a word processor.

That wasn't until a few years later. And that was because PCs became popular. PCs became popular because I think one of them is people involved in money like the spreadsheet. They said, oh, these PCs are useful.

And then when they were good for word processing, the journalist said, oh, PCs can be useful. And then people, their kids went away to college, and the only way to communicate with them was by instant messaging and email. Oh, computers can be useful when connected to a dial -in remote stuff. And then the internet.

Every time there is some new capability from a hardware viewpoint, or hardware and software combination, that is much, much, at least 100 times better than what was before. Like, think the mouse is so much better than arrow keys, right? In terms of precise positioning quickly. Every time the laser printer compared to the typewriter output, you know.

It was a ball, remember, the Selectrix. I don't know if people know that. But in the line printers, the laser printer could put 300 dots per inch each direction and do pictures and everything. It was so much better than what came before.

That opens up new capabilities. So you then say, with this new capability that is two orders of magnitude, 100 times or more than what I had before, what does this open up and what is it good for? This is good for something I used to do or I needed to do, but didn't do. And that then lets us move forward.

Portability makes a big difference. It's a big difference to be able to find, you know, to find messages in a fraction of a second compared to have to dial up, you know, to AOL or something and log in to read your emails and stuff like that. Or, or, but that was so much better than waiting for the mail to come, you know, the snail mail. and each of those things, the fax machine, it opened up so many things, right?

Because you could send a document, anything drawn or typed or whatever, you could send it immediately to other people. Now, what was interesting is about putting up with things that aren't so good. We worked really hard in word processing to be letter perfect, they called it. The output, you wanted it to look as good as a good typewriter.

Okay, really really good. Then we had the laser printer that could produce fonts of all sorts, perfectly sharp, indistinguishable from the old stuff. Then what do we do? We said, oh, we'll stick it in the fax machine that turns it down to 150 dots per inch and you know, whatever and it's only black and white and no no gray and You know, we'll send it that way We'll take our litter perfect thing that we work so hard and turn it into this barely readable thing on paper That was kind of wet and you know faded whatever but we could get it immediately right and you know And we put up with that when we first had cell phones Can you hear me now?

Can you hear me now? It was, you know, was, but. And they were enormous. They were huge.

They were huge. But at least you could carry it with you. Right. And so that's, you know, you look at when you have a new capability, it opens up some things and you're willing to put up with it not being as good.

A lot of this is described by Clay Christensen. May he rest in peace. And the innovator's dilemma. Yes.

So about that whole thing about when you have something that disruptive technology. So we can look at that for any new technology we have today. And how is it better or not from what we had before? And what is it good for?

What is it not good for? And what doesn't it matter? Right, right. This sort of trade off when there's something like some new capability, but then there's this trade off between quality and then port, like portability.

I think about music also, right? It's like the quality of music by changing it and having it be into an mp3, you're losing a lot of quality. For those that can hear that, not all of us. The thing is that the portability and the fact that you could carry in your pocket hundreds of songs, you know, to listen anywhere.

But the quality was pretty you know compared to a transistor radio You know listening on the radio waiting to hear whatever happened to be able this tinny little speaker when you had those really good earphones that The Walkman when I when Sony did the Walkman they did these really good earphones with these special magnets and whatever So it actually sounded better then it probably would be if you used your expensive stereo in a noisy room so and it doesn't matter because you're you're portable you can do it when walking or running or when you're commuting so yeah this this trade -off of uh and then eventually they get better where the sound quality you have with the latest little earphones wherever running off of you know directly off your watch or your phone or whatever are like concert quality compared to our old speakers that we had you know in our dorm rooms or something like that years before for an expensive stereo.

So it does catch up that's the innovator that's exactly Clay Christensen's about how eventually the toy becomes fixes the differences between it and eventually becomes better because of the volume and stuff like that. So going into VisiCalc and creating something like that, what were, I guess, a couple of questions. Like, so when you're creating this general purpose tool, first, it's hard to think of all of the possibilities that you could, you know, that you could hit, right?

But I guess could you talk about... I had no clue. I mean, I knew about a lot because I was at business school learning about all different aspects of business. So I was exposed to many uses.

Right. but it was a fraction of the uses that it actually became used for, even in the first year. Right, right. Yeah, your story about...

So you have a great TED Talk that I encourage all listeners to listen to about the history of VisiCalc, but you mentioned there that you used the spreadsheet basically to help you with your case studies and to do projections and to do that... those sort of things, which personally was, is kind of cool for me because one of my first case studies was, was yours, was your case. I mean, they're just, it's, there's so much echoing of the past. Yeah.

So it's, it's, it's, it's a pleasure to be. And it was highly interact. VisiCalc is highly interactive. I worked on the word processor with somebody in the name of Jack Gilmore, may he rest in peace, who was on the word processor.

He developed what is arguably one of the first video games. Well, it is one of the first, but it really could be the first video games in 1951. Oh, wow. You know, and, you know, on the whirlwind computer.

So, you know, this moving along and from person to person to person is kind of cool. Yeah, it's it. Yeah, it's it's it's really cool. So while you were developing it at like, what were the challenges that you were facing?

Were there like? I mean, I guess the technical challenges of trying to fit it. I know you mentioned that there was a particular upgrade in the Apple II that made it possible, I think, or there was some additional memory or something that allowed it. Oh, well, I mean, there was stuff.

The thing is that, well, the Apple II was good enough, but the challenges were the Apple II had a very small screen in terms of characters. 25 rows of 40 characters wide on the screen for character based. And it had two arrow keys. That's it.

It had paddles you could play a game with, but I was gonna use that like a mouse, but I gave up on that very early on. Like the... No, no, no. The type you turn.

Oh, okay. The turning one, yeah. I should go get it. I have one.

An old Apple II with it. It was for playing punk. Right. You know, that was that type of stuff.

Or Space Invaders, which is what everybody wanted to play, which was a game on the Apple II. So the challenge was you're working with data that probably... You know, you're going to want a lot of it. So luckily from my word processing background, I was used to scrolling and dealing with scrolling, but how do you deal with scrolling on such a small screen?

So we had to do things like what we now call locked pains. We call them titles where you would have, you know, the name, some kind of maybe a row names or something. And as you scroll, you know, they would stay on the screen. Right.

and I'll let you have two windows so you could look at two places at once. You could do this and see the results of a sum somewhere else. And if you did two scrolling, they would be synchronized. So we decided to put that software in.

Help, we wanted to have a help system, but there was no space to put that much information. The only help we had is you typed a slash and it showed you what characters you could type for the commands to put in. And slash IR is insert row, which I think still if you type slash IR into some versions of Excel, it still inserts a row. Those letters that we showed on the screen were the same letters that we were comparing what you type to, to see if, you know, what to do next.

I mean, to save memory. So memory, how do you store the numbers? First of all, we had to make it that if you saved it and bring it back again, it would still fit. So if you used up every byte of memory available on your computer, there was no putting it out to disk or any of that paging stuff that we knew from our other day.

So you had to build it so that it would be repeatable. And therefore you needed certain things to be very quickly garbage collected using fixed size, whatever. So what was the largest number that we would store? Because we needed to know that.

and Bob and I talked about it, Bob Frankston, who wrote most of the code, and we'd said, okay, let's imagine that people would use this to calculate the budget of the United States of America in dollars. That's enough significant digits before we go to scientific notation, where it's powers, you know. And sure enough, It ended up being used for that a few years later. But we had to make decisions like that.

How many columns, how many rows, max, because, and when do we allocate memory to be fast versus how do we do it so that you could actually have a spreadsheet that could be this wide or could be this high, but it couldn't be both at the same time because there wasn't enough memory for it. We only had, you know, I don't know, like 30, you know, 30K thousand bytes of memory for your whole spreadsheet. I mean, we started on a system that was 32K bytes for the whole system. That was the program, the operating system, the screen buffer of what was going to be on the screen, and your data.

We could fit a few K of data for that. Most people had 48K, which was the biggest version then that we could use if you... bought a special card to run some other program that sold this card, you got another 16K of memory and we could use that card to give you more memory and people would spend hundreds of dollars on this card and throw away the software that came with it in order to get the extra memory for bigger spreadsheets. So we were very constrained.

We had a program, Bob had a program really, really well to be able to scroll and, you know, display. turn it all into the screen. You had to turn it into the screen dynamically as you scrolled. That's a lot of programming.

It has to be fast enough for the repeat key. What we didn't know is that the repeat key was done with hardware that isn't based on time, but was based on an RC circuit, which means it varied from computer to computer. Our computer, you could scroll and stop, and it would stop. But some, if you stopped, it still had a few extra keys and kept going.

But it had to be that fast. That was a real challenge to make it fit in memory, do all the things that we wanted it to do. And how do we decide what not to do? We wanted a great help system.

We wanted more functions. We wanted, you know, I mean, you're just, you know, it's deciding what not to do is a very tough thing. And we had to make it easy to understand, you know, to learn. And people learn by using a reference card and a manual.

They either read the manual or they looked at the reference card or both that tell them, if you push this key, it does that. And we had to make that as simple as possible and we succeeded. Yeah. And useful.

Yeah, for sure. I mean, it's still the most used, one of the most used software. So VisiCalc was the precursor to Lotus 123, precursor to Excel. And then Google Docs and dozens and dozens and dozens of other spreadsheets that have been built.

Right. Linux has their own open source version as well. You know, what's funny is, you know, designing software, often the competitor is the status quo. So you created something that's like the status quo, like any...

No, the status quo was paper and pencil and a calculator. I'm saying that it became now that that has... No, it was a status quo. It was pretty easy to beat us out if you were on a more powerful computer because you had more columns.

You had graphing ability on the screen. You had lots more memory. And if you happen to know what people wanted, things we left out like commas in numbers or the thousand separators, the columns being different widths and some other things. And there were certain things we left out because they were hard to program or we weren't willing to make certain compromises that other products like 123 did do and did very well because they had those.

And then when there was new hardware, so Excel could take advantage of fonts. Just think what multi -plan from Microsoft put in and then Excel used, which was the ability to have a single line between each cell, to have a grid visible. That's a huge advance for many applications. And they were able to do that because the hardware, initially of the Macintosh, the early Macintosh that they did, let you do that.

So there are all these advances that you could put in new products. With Excel, you could run it under Windows. And it came bundled with your word processor with Office. And so whenever you bought your computer, it came with it.

So that was a reason that people would go with that. So 123 was just much better for the IBM PC. It was tuned so much for it. was so tuned to the IBM PC, which had just come out, was now available from a major manufacturer.

People now knew PCs were useful. So they were finally all buying them. So Jeff Moore's, you know, inside the tornado and about the early adopters now going to the early majority where there's a huge uptick. And IBM PC was the computer of the time.

And the spreadsheet of the time was one, two, three. And it came even with a piece of plastic, molded plastic that went over the keyboard's function keys to label them. So you knew where help was and you knew where this was and all that because you had physical labels on this piece of plastic tied to the hardware so much. So why not buy that?

And so what did 123 cost like? for $500, the computer with the printer and all that cost you several thousand, like three, four, five thousand dollars perhaps for your computer with a printer and all the things that you, the screens and the stuff that you wanted. So the software, who cares? But you bought it for the software, which is what it means to be a killer app.

That it's worth buying the hardware for that software. Yeah. There you go. But thanks for tying that back in.

Yeah. So the idea of having that killer app. The matter is when you already have the hardware. Right.

So like when Netscape came in with a nice browser, the idea of that type of a browser, there had been browsers before, but they weren't integrated with screen, with pictures and stuff the way that it was. There were just a few and it was available for free. All you needed was a personal computer running Windows. and you needed to have that computer on a desk and you wanted it connected to the local area network to be able or somehow connected to what could be the internet or whatever.

You needed that hardware all in place. And companies had that because of Lotus Notes, a product that Lotus did that was really good for email and workflow and all sorts of stuff. And it was one of the premier applications for Windows to help Windows being developed. And so the hardware was already there.

You didn't need a killer app. All you needed was the good app. You don't have to buy the hardware if it was already there. And then, um, when Napster came around, you know, you don't need to buy anything.

Right. When, now when you needed a music player is when the MP3 stuff, you know, and Napster and whatever came in, that was, you sort of needed that. Uh, if you wanted portable music. But when, you know, when Facebook came in, they already, students already had computers.

They already were connected to whatever. So they, they didn't have to be a killer app that way. They just needed to be a useful app. Right.

Almost like use the system that's in place. to then kind of to move things forward. Once you have a fertile environment, if you want a fertile environment, meaning it's a thing where it's a good place to do different types of applications and stuff that lends itself to innovating in different areas. So the PC was a fertile platform of sorts.

And once you have that, you can do things. for that. And we're seeing that today with AI, where people are using AI systems that other build as the fertile platform in which to build something else, now that I have something that can take inputs and then do outputs in English. In terms of win something, it's hard for adoption versus easy for adoption.

Right. Now we have adoption in... Just a thing I saw that David Smith wrote. David Smith wrote a product called Widget Smith.

which is on the Apple iPhone that lets you put widgets, customize them. And he writes about it, how he was building it originally for the watch and then he was into customization and Apple made it possible to do that. And then he came out with it and it started selling like normal. Then somebody put up something on TikTok where they said, boy, you can customize exactly what I want.

Customize my screen to be exactly what I want. Within hours he became the number one app for a week or for two weeks on the Apple iPhone Right. He sold millions and millions. I think what do you say a hundred million copies or something didn't sell them all But he you know, they're free.

It was for free, but you can buy the the upgrades You know, I mean within within days or within hours today is quite different Right how fast things how fast you can go up and how fast you can go down. Yeah That's for sure. I wonder, are there any new criteria that you add in to making a killer app today? I haven't thought about it much, but I think the whole thing of whenever there's a new capability, look at what is it good for and better than anything before by not a little bit, but by a lot, 100 times better type of thing.

Right. Now that you have that, what does that enable You know, so let's take machine learning. Yeah, okay the old machine learning It was used for visual stuff to recognize things and really kind of cool on the production line You know, you could see is this a good one or a bad one and you could then automatically You know do some quality control and stuff Well, what happens if you can put it on a small enough ship to be able to detect people by eyes? Being able to say that's a face All you have to do is figure.

Well, that's a really tough thing to find faces. OK, but we figured out machine learning and stuff. They figured out how to do that. And they put it into cameras.

And we take this for granted. The fact that a camera can figure out it's a face and then do the autofocus there, not just the dot in the middle, wherever you point it used to be wherever you pointed it. That was the cool thing. Right.

Where you pointed it was where it focused this. It's able to say, oh, I see a face. I see two faces. Let's try to figure that out.

What was it good for? It was good for that. It was good for lots of other things. So what is it good for and that we could do that we couldn't do before that's worth it, that pays for itself almost immediately?

I say within two weeks type of thing. When you got desktop publishing, a laser printer, something like PageMaker or later Word and stuff like that, being able to... get your output and, you know, do a newsletter or whatever. The old days, you had to type it up, send it out to a typeset, you know, to a person who would typeset it and then print it to the printer and come back.

It paid for itself the first time you used it. There was a two -week payback for a lot of these things. VisiCalc paid for the Apple II. You know, there were certain...

Programs you could use to do financial forecasting not as easy to use and whatever but for your particular application You could use a time -sharing system six hundred six thousand dollars you a month. Well for six thousand dollars That's much more than I to get the top end. You'd get an Apple to a nice monitor Sony's Trinitron probably or something that I could use to watch TV and a printer the best printer that I could get at the time, you know, a daisy wheel printer or something.

So it pays for itself the first time I use it in a couple of weeks. No -brainer. I mean, it's obvious. None of this one -year payback, two -year payback.

So, you know, cell phones, the first time you're stuck somewhere and can't, with a flat tire that you can't change, in the middle of nowhere, and you find out you could have had a cell phone you know when your mother does that you buy her a cell phone right away i did that you know she had a flat tire and you know whatever i said mom you're getting a cell phone right this is many many years ago um when but or late enough for cell phones so that's um you know when is it obvious that it pays for itself the first time you really need it or something so those that sort of gets into today's stuff um and separately is user interface.

Why was my spreadsheet implementation versus other calculating things that came out? Why was mine so successful? Was it just the marketing? Was it just the cool name?

Was it our implementation? Was it, you know, I mean, there were other, you know, was it because it was on the Apple II? I mean, why, what particularly, what was the things that made it, you know, but a lot of it, is user interface and stuff, and what were they good for, and is it fun? In the early days, when I first did computerized typesetting, in our office we had a typesetter, thing used chemicals and stuff, whatever that you could use as an output, like a printer.

And out would come this paper, little shiny paper, that you could do fonts. you know, Times Roman and then Italic and all that stuff and different point sizes. And the first things we did were look like ransom notes. These are our status reports, you know, where we could have just made it look like a typewriter, but no, we wanted to do all that.

And sure enough, when the first word processors came out that did fonts with laser printers, what did everybody do? They went overboard. They use it for everything, you know, and they made fonts like, you know, and it was the most garish thing. And now we've sort of settled down.

Right. The same thing. It's the hammer. Once you have a hammer, everything looks like a nail.

I'll use I love this new thing. Let's use it for this. Let's use it. I don't know what it's really good for, but I'll use it for everything.

Right. With today's stuff, what's it good for? You know, I can ask it a question, it gives me an answer. That's good.

Well, it turns out it's not very good for certain answers. Right. And it's really good for certain other answers, you know, and certain type of questions. And understanding that is important.

Yeah, 100%. And I think the challenge right now, I mean, you've seen all of the cycles, you know... Lots of cycles. It goes back, you know, back to the Stone Age.

Right. The AI winters now... Look at my wheel. Now we're in this, people argue where we are in terms of the, you know, the hype cycle for AI right now.

I think we're really high. I think some people are getting into that trough of disillusionment also a little bit, but I wonder for you seeing it from your perspective, how do you separate kind of the hype of AI and the reality of it? Well, a lot of people hyping it don't know what it is or what its limitations are. Right.

What I used to say, let's see, when this account first came out, you could explain it or show it to people, okay? And if you showed it to a computer person, they look at it and say, what's so special here? I can write a program that does that, you know, and I can do better and stuff like that. You show it to a regular person, a normal person, they say, computers can do anything.

They can forecast the weather. They can tell you whatever they do. What's so special about that? But if you showed it to people who actually did that stuff, they would start shaking and say, here's my credit card, please take it, give me that, give me that, give me that.

So the same thing's happening with a lot of stuff with like AI today, is people look at it and they say, they don't really understand its limitations or what's what, but it looks like it's real smart, like a person, it can do things I can't, and I can ask a question because of an answer. And it sounds like a person doing that who is that confident. must be right and it's a computer you know and it's trained one on the whole web and the web must have everything there must be everything on the web well we know that the not everything is on the web and not everybody puts stuff on the web so it's not trained on everything and the people who are training it to check it don't always know the right answer right um etc for me what i need in my application stuff i just saw with today or something that a lot of training was on certain populations of medical stuff is in the literature because they only did the studies on certain people they could get to.

Right. But that's normal people. I mean, you should understand the, you know, well, maybe from my experience, you know, it's different that the training now is based on that as that's what it is. There's an old joke.

about the policeman sees this drunk and he's looking around on the ground and he says, what's the matter? And he said, well, I lost my wallet at 23rd and Walnut. He said, well, oh, well, we're at 19th and Walnut. Why are you looking here?

He said, well, the light's better. And so that is that when we have the data, we use that, whatever we can get. Whatever is easy to measure is the thing that we measure even though that may not be a good measure but there are people who don't know that who just assume that it must be all -knowing and It's better than me. So therefore it's good.

So so we learn people who really use it learn to use the tool, like those that learn to use the spreadsheet, those that learn to use the word processor, those that learn how to drive a car versus, if you don't know how to drive a car, you're gonna crash and you might kill yourself. Once you've learned how to drive a car and stuff like that, then you can do also, and if you're really good, you might be a race car driver or something and can do it, but you have to learn to do that.

dancing, you know, learning how to dance. You know, some people are naturals and she goes, look at one other person. The same thing about these things. We have to learn how to use them and what they're good for.

Unfortunately, they keep changing. So I become a prompt master. I'm really good at prompting for GPT -4 for three. And I know all of what is good and I know the way to do it and how to, you know, and then of course it's changed.

and suddenly my skills may not be as useful, things that I worked really, and maybe there are different ways to convince it for this, and that's just with one type of input, with one type of output, and it's changing so fast. That's not a recipe to depend upon it. it's great for advancing. Are you going to depend, let your life depend upon it?

Reporters would ask lots of questions and learn lots of stuff about an air, it's something and report about it. But if you knew that particular thing, you were at the event and you knew whatever, you knew what they wrote isn't exactly right. Close, but not necessarily exactly right. Especially if they weren't, you know, had been in that area and whatever.

There's the same type of stuff here that we have something that, yes, it's pretty good. Is it perfect? Well, right now we're learning like encoding. We're doing vibe coding, whatever.

You have to check what it does. Yeah. Well, if you don't know how to check, you know, if you don't know the type of failures that you might run, and we don't even know the type of mistakes that it might be prone to, that we have to look for what type of failures and what uses will be used for. And it keeps changing.

So what it was good for, how do we know if it's still good at the thing that I want to do? It's going to take time. to settle these things out. Is it still better than not using it?

For some things, it is. And some people who know how to use it, like any tool, can get great advantage. Anybody who had a spreadsheet, suddenly, in the early days of the Apple II, and nobody else knew they had it or what it was, thought they were... these geniuses you know they unbelieve you did you know what people would do is they would get a job quote it for this many hours of money go away and then eventually when they needed to do it spend the 15 minutes use the spreadsheet to do the answer and then come up with it and wow whatever i got my money's worth um you know we're we're in that that world for certain things right so we'll see um and what was the best way my idea of uh using the paddles was not a good idea on the Apple II.

The mouse is good for certain things on the spreadsheet, but the arrow keys are good for others. Using the keyboard is better than reading into voice input for a spreadsheet with numbers and words and all that. You can tell by anybody's texting who texts you voice. You know whether, do you want to depend on your doctor using that type of stuff where the difference of voice recognition is life and death?

You may want different type of inputs with the right type of feedback. The feedback that we get from today's systems sometimes is, well, I can read it and I can tell where it's good or bad, or something that doesn't matter because anything that gets me in the ballpark. In other words, I'm trying to get I'm trying to get within the ballpark. I don't have to get to the seat.

I need to be able to get to the ballpark. So therefore, anything that gets me to the ballpark or nearby most of the time is a help. And you have to understand that. I mean, can you imagine if with today's systems at the error rate in terms of even small or big, you're using it to get to your seat in a ballpark or anything else.

You think a lot of us would end up in the same seat by mistake and not in our others with, you know, I mean, we're talking about in a stadium, if you have 0 .0001 mistakes, you're still going to have some people who are in the wrong seats. So you have to have the error correction and stuff. So I don't know how we're going to do that.

I know that there's a real challenge with the interfaces to these different models that are using the technology that the LLMs use. and the type of training and whatever, what is the best use of this technology for what type of things? I'm really excited by new hardware that's coming out that will be able to do it apparently. Some of the new phones and stuff, the particular chips in them are powerful enough and coming with the right software to be able to do certain things even on...

in your hand and what's that gonna open up? We're not gonna just be talking. You're not gonna wanna talk to it. You're in a noisy environment.

You're gonna wanna be able to do other ways of interfacing. We're learning. Remember with a lot of the visual stuff, it uses hand motion. Well, is that really the best way or is it better to have a controller?

We now move back to controllers. Remember the iPad? Right, the touch screen. Yeah, with no pen.

Well, now you sort of must have a pen to be able to use some of the stuff on the iPad for some of the applications. But we wouldn't have a pen when we can do touch. But it turns out that the interface is better when you have the sharp point. and the ability that people have, but you also need others for those that can't use their hands that way.

There's accessibility. There's where I'm using it, noisy environments. Where I can't, how do I present things to you so you know what to check? Write me a 10 ,000 word treatise on such and such.

Well, how am I gonna check if it's right? How do I know where to check? I mean, it's the teacher will just use an LLM to tell you if it's right or not. No, I'm just, you know, and what's that trained on, you know, what, you know, what class is it then?

And do you trust the person who trained the train, you know, to train those systems? There are people you can read about the people. They get some pretty good people to train these systems, you know. to say no, you know, no, LLM, you got it wrong, you know, whatever, and it fixes it and stuff.

They still do weird stuff, you know, weird stuff. I think it's that's one of the flaws, actually, that we're using human preferences. So it's. Trained to output things that humans prefer and that's sometimes just off of like what you see and what you look like and sometimes the formatting of an answer can cause you to like it more than some other answer and the fact how Factual it really is isn't always checked.

There's so many there's so many challenges for it We're so early, but it's interesting to see the way the pendulum swings It's interesting to see hat what the new interfaces are gonna are gonna be like for it. Yeah Um, you raised so many good, do we need new hardware? Do we knew? Well, first of all, we knew, we knew we needed new hardware to do the calculation.

So therefore they've been putting the neural engines and all that, and all these hardware. That's okay. So we've done that, that at least, but is there other hardware? Like we needed bitmap displays, meaning you could, you could address every dot on the screen in order to be able to do, you know, uh, mouse and stuff like that.

What other hardware, what interfaces, input -output devices, do we need for the particular applications for which these things are useful for? Like if you don't have an autofocus motor, what good is being able to recognize where the person isn't focusing, right? It'll be interesting to see how those co -evolve with each other as these new capabilities, you know, arise and emerge and then there's going to be new hardware. I don't think we know yet what the right interface is exactly to take advantage of all of these things.

But it is interesting for each of these things, which interfaces for each of these things. Right. Is it going to be right for we still know people still buy laptops even though they have just as powerful phones. Why?

Well, there must be something about that laptop that is better for them that they're willing to spend a thousand bucks. or a few hundred to 2000 bucks for that laptop versus their phone that they're also spending hundreds of dollars or a thousand dollars for. And so we have to sort of understand that whole thing. Yeah, yeah.

You raise so many interesting points, so many things to think about. It's hard to be an engineer because there's so many questions. And how do you decide, now's the time for me to bet on this? This combination, is it too early or is it too late?

Am I choosing the wrong combinations, the wrong application? And then there's the old market fit, do people care? And a lot we don't know. That's why it was interesting to see David Smith, what he wrote about how he...

He didn't know that there were I mean he cared about fonts particular you choosing this font and this color and this type of little layout whatever the style for a clock or the showing the weather because you cared about He cared about that, but he didn't know that there were a lot of people that did care He thought most people didn't because on other products they didn't care But it turns out on your home screen. There are people who view the home screen like they view the case, right?

and they view their clothes or something like that, that that does matter to them. And that's how they express themselves. He happened to hit the right combination that others did not. They also had some stuff, but they didn't allow you to do the chartreuse that you wanted or something like that that made all the difference in the world to some people.

And that was good for them. That's true of a lot of you don't know the space. You're not that type of surgeon. You may not know the needs for that type of surgery.

If you're not that type of business person, you may not know the needs of that. Oh, the person using it works in this environment, physical environment. What must that hardware have? They only have one hand to do because the other hand is busy.

So they have to do one -handed operation. There are just so many things. That you and you don't always know it and they have to go right for you to be successful. Yeah, it's a good point I mean, yeah taking into account When you're trying to design anything right, you know who your end user is what what environment they're gonna be using you're Offering in the context that they're in the experience that they've had in trying to create something that can kind of address You know all of those things.

That's why I think that Robots, you know humanoidish robots might be interesting in that So much of our society in the world around us is built around a person within a certain range with one or two hands, you know, with this many fingers and whatever, and with this ability to move, et cetera, and this strength, you know, and speed that therefore... maybe those input -output devices, which is the humanoid robot, may be appropriate to fit in situations that we couldn't automate without specialized machines.

It's a real challenge, but what are the physical configurations that are appropriate for our physical world? You know, the fact that a cell phone fits in your pocket matters when the first scientific, really good scientific calculator came out, the Hewlett -Packard calculator, HP 35. The story is that Packard, Hewlett -Packard came to his engineers and said, I want something that fits in my pocket, my shirt pocket. They took out their measurement and they measured his pocket.

And sure enough, it could fit. It was, you know, a certain thing that he could fit in his pocket because he needed to be able to carry it around, you know. Um, before that we had slide rules and we had special strap, leather straps that fit them on our, on our belts. And that wasn't for regular people.

This you wanted to, and that size calculator became dominant. Right. And same thing with the phone. It fits in the pocket ones that were too big, weren't right at first.

Right. Once they, once the, um, like the Nokia, uh, small cell phones came out that you could fit in a purse. and you can fit in your pocket. That fitting in a purse and fitting in a pocket make a difference.

What's really nice about the new iPhone Air, it's really, really thin and fits in your purse a lot better, among other things, which is packed. You put your phone right. Right. I mean, going back to what we were talking about earlier in terms of like how the pendulum swings back and forth.

Yeah, like cell phones were smaller at one point in some ways, and they now have gotten bigger and now they're going back to being a little bit smaller and thinner. And then there's other tradeoffs. Now we're doing folding together. Right now there's the back to the flip.

And then there was a time when who would ever have a full touch screen. And then there's other tradeoffs in terms of when you make something so small yet so interactive. right, there's trade -offs in terms of battery life, right, and things like that, and then you have to weigh, you know, those trade -offs, so that goes into... What's the sweet spot?

We don't know what the sweet spot's going to be, and luckily, lots of people are trying, and we're seeing what sticks. Right, right. Which I guess is just the exciting thing about technology and evolution. Always been.

Yeah. you know, look what happened with automobiles, you know, and stuff, you know, automatic transmission, you know, and stuff like that, you know, well, anti -lock brakes and whatever. And then something's caught on and then it became standard. You know, I don't know how, how to drive a, you know, an older car that doesn't have the backup checking and a backup camera and whatever, you know.

And then we become so used to it, you know, then you start to become so used to it. Yeah. Which, which is. That's why I'm really scared in my car that doesn't have those.

You know, it has a backup camera, but one of my cars doesn't have the sensors. Right. And now we're so used to hearing that sensor when you turn the blinker on that, you know, that if someone's in your thing, I, you just always need to check, especially as you get your neck back and all. So coming towards the end, I have two questions that I like to ask.

What advice would you give your younger self? Maybe when you were kind of just say starting your career. Oh the career advice as opposed to personal advice Well, you could do both but personal advice is yes. She's the right one Okay, so that's you know, okay that's very important finding the right that's very important very important a finding a partner is is You know important and so what advice what would I want to do the different?

I mean because I I really like how things turned out I mean, things could have turned out different where I could have made lots and lots of money, which I did not because of various business questions and stuff like that. On the other hand, I like my life and it's better than I expected it to be. So I don't know what advice that would change what I would have done except that, let's see, I mean, choosing the right, you know, right horses to get on. I mean, that's the, this, which technology one to ride, when, how to make that decision.

But I still don't know. I mean, all the things I've told you. Some of it comes from what I've said way. I was saying back then that I learned then that's kind of that's kind of tough You know family matters enjoy Enjoy the good times because they may be the good times There's movie as good as it gets there are some things that are as good as it gets enjoy them when you have them When you have the really good time enjoy that don't feel bad that it could be or whatever.

I've had some really Special things happen in my life, you know and been at unusual situations and no longer have that So the more that hopefully I enjoyed them at the time that I could enjoy them, that's good. Right. There's like a great quote that you reminded me of, of it's like about the good old days. Right.

And sometimes you don't know, like I wish there was a way to know you're in the good old days before you've actually left them. It's from the office. It's a good quote that I recently came across again. Now, many, many years ago, early in the days of Apple, Steve Jobs came to speak at the Harvard Business School to an entrepreneurship club or something.

And I went to go visit. I was a graduate at the time. I had graduated at the time. So I went back and I raised my hand and said, hey Steve, you know, where are you, what are you expecting for the future?

Where are you trying to go? And Steve said, it's the journey. He was not looking for where he was. It was the journey.

And that, you know, kind of, oh. You know, I had asked the same thing Gates at some point and, you know, Gates had, I'd asked him about some stuff and he talked about how he was going to give money away. And he sure did, you know, that, you know, this was his success. Look, we'll look in the future and if he's successful, he'll, uh, you know, he'll, he'll get back that way.

So that was the thing about enjoy the journey. I always, when I did accompany What I did was when I was paying people and sending stuff I said I want to do it that when you look back if we if we fail I want you not to Feel bad about that. You wasted your time here. I wanted you to say, you know, I wish I hadn't You know, I want to make sure that I paid you enough that it isn't costing you to to become poor or something like that I wouldn't that was now that may not be the best business But I was a feeling I had that I wanted you want to do it so that you won't regret That you were doing that stuff because of what doesn't if it doesn't pan out from that viewpoint so do things that's the thing of doing it what you love and You can't always follow that.

You don't always have that option. But if you can that's pretty good Yeah, it reminds me of your earlier remark. Life's a journey, not a destination. We might be seeking all of these things, but it's really about the journey that gets us to where we are.

I'm afraid that this last question is too close to that last one, but I always ask it, so I have to. What has a career in tech taught you about life? I love tech, and I love that. And I'm happy that I did it.

So I'm happy I chose that and that I'm still doing it. I'm still programming. You know, I'm still writing and I like, I'm still programming a lot by hand. You know, cause I like, it's like people who like to paint.

They're not Picasso. They're not Rembrandt, but they like to paint. Even though they're not, you know, they're people like to play basketball. They're not LeBron, but they still like to play.

You don't have to be the best. So that's, that's one of the things about, you know, about your career. But you also should think about what you're good at and what you like. You know, there's these retrospective I saw last night about Robert Redford and about what he went through figuring out what his calling was in life, where, you know, he realized he was in arts and he wanted, he thought he was going to be a painter.

to france and stuff and he thought he was going to be a painter but then he ended up you know in school for for drama and stuff like that and a teacher who told him you're you're good to stay in that you know type of stuff and then he did he ended up on stage and then he ended up in one film and then he ended up behind the camera you know as a director and figuring out what and then then of course Sundance in terms of giving it back in terms of his love of The outdoor of of outdoors and stuff like that.

That's an interesting thing to think about Look at people's paths how they got where they got especially ones that seem happy with the path that they took and Why and so that's a good reason to read biographies of various people that have done stuff to talk to people Who would not have written a biography or there being a biography of them? To find out and about that somebody contacted me that I knew from, from high school and said, Dan, I hear that you took pictures back in the old days and you have pictures of the play I was in.

And so that's, you know, yeah. Um, and you still have them. I said, yeah, I scanned them for one of the other people who was his kids, didn't believe that he was ever in a play and I was the lead. And so sure enough, so I had the pictures and this turned out to be a major memory in her life.

I was into photography. And I had pictures, I had great pictures of her. And I found out something I didn't know. She was a shy person and she had a crush on a guy.

And she got Maria and he was Tony. And they, boy, I have pictures. You got to live out a fantasy. You get to live out all sorts of fantasies in life sometimes that you don't expect.

I mean, they're not married and they haven't, you know, they're friends, but they, you know, they're different parts of the world, whatever. But looking at that piece of her life. And that turned out to be a major, in hindsight for the two of them, that play turned out to be, in high school, turned out to be a major good point along the way. I'm sure in terms of other things in their lives, they've had some other high points.

I mean, I'm a grandparent and there are just such incredible high points you can get as a grandparent, and stuff like that. And being a parent and the same thing, So enjoy those and they will happen and don't belittle them when you have them. So that's, that's advice that I have, including in your career. You know, I mean, even if you're going to get fired at some point, you may have, you got to speak in front of a huge crowd and they clapped because they liked what you said and you liked that feeling.

Right. Maybe, I don't know. Or you have a product, you built a product and people actually use it. And then you get I did a product where you draw on the screen, but you can draw big and it shrinks it down.

You could mark up PDFs, okay? So on an Apple, on an iPad, even though you didn't have to write real small. So I hear from this mother who said, thank God, I really love your product. My daughter has cerebral palsy and now she can do her homework herself because she's able to draw big.

on the homework and she doesn't have to tell me what to do. Thank you. I mean, that was like, I still remember, it was years ago. I still remember that moment.

So you never know. There's a line in the play rent, you know, about helping other people and stuff like, I mean, and you know, this thing about the relationship with other people, you know, is part of what it's all about. So enjoy those. If you do, there may be other things you enjoy more, enjoy them.

Ice cream. I think it's really good advice and it reminds people basically, like, you never know what the consequences of your actions are, but if you're kind of doing it and you have the right intentions and you're doing things for the right reasons, you could make an unbelievable, you know, you can have an unbelievable impact on other people. I think that, yeah, that's really beautiful. The speech that I give that the TED Talk is a piece of, I've had people years later say that I was inspired by that and I went and did this thing totally different and whatever, and this product that we all know about is there because, thank you.

Yeah, it's amazing. Yeah, that's incredible. Or, thank you for the spreadsheet. I was helping this other person learn the spreadsheet because they needed it, and now they're my husband.

Thank you very much. Yeah, and you never know. You never know. You never know what could happen because, yeah, because doing one particular thing could then lead to unbelievable impacts and unbelievable consequences.

For others or for you. Yeah. Yeah. I think that's a really good place to wrap up.

Dan, this was such a pleasure. Same here. I appreciate it. I love your questions, the areas where you're going after and stuff like that, and knowing your background makes it even more so in terms of your technical background and stuff.

It was really, really a pleasure to have you on. Thank you for all of the incredible work that you've done. Yeah, your TED Talk is inspiring. It's inspired me.

Your case study was one of those first ones. Reading that and, you know, connecting with you is a real honor and a real pleasure. I really appreciate you taking the time to chat. Let me pick in your mind.

My pleasure, too. Thank you, Dan. the first electronic spreadsheet, and the killer app that helped launch the PC revolution. We explored what five decades of platform shifts teach us about today's AI moment.

Dan's framework is simple but powerful. Breakthrough innovations must be a hundred times better, not incrementally better. The same questions he asks about spreadsheets and word processors apply to AI today. What is this genuinely better at?

What does it enable? What trade -offs will people accept? Does it pay for itself immediately? Most importantly, Dan reminded us that we never fully know the impact of what we build.

Whether it's a mother whose daughter with cerebral palsy can finally do her own homework, or a couple who met learning spreadsheets. The moments worth remembering aren't the product launches or exits. They're the unexpected times. when your work changes someone's life in ways you never imagined.

Thank you for listening. Be sure to subscribe and share with a friend or colleague. Until next time, keep on learning.