Hi guys, I think we're the finale of the day, is that right? I think so, we're getting towards the end. Well, it's a full room, so thank you guys. This is such a great panel and such a broad topic, I feel like we could go on for days. We have about 30 minutes, so we'll try to get through as much as possible. You guys should introduce yourselves because you go way back and sort of had a baton passing in your career to Twitter. Jeff, maybe we'll start with you. You can sort of introduce yourself to everybody and you also have some news today which we should get to.
Yes, thank you so much for having me. Great to be here, great to see everyone. I'm Jeff Seibert, I'm the founder and CEO of Digits. This is my third company. Previously did a company in the document collaboration space, mobile space, sold that to Twitter. And today we just launched Digits. Actually, we've announced the first end-to-end accounting platform of the AI era. There you go. So you can think of it as a complete replacement to QuickBooks: invoicing, bill pay, accounting. And the big advantage is unlike all the manual bookkeeping accounting workflows you're probably used to, Digits automates 93% of it. All your bookkeeping is done for you. So it's been really, really exciting.
Extensively. 18 different models in production, orchestrated as a symphony to actually do the accounting workflows.
We do, yes. So yeah, the way to think of it is we use GPT as sort of our base tier and then we do a whole ton of predictive models all in-house, custom trained on top that really understand double-entry accounting. And so that's how we can layer in all the expertise of deep accounting knowledge with the general knowledge of OpenAI.
Kevin, what's OpenAI? This tiny startup most haven't heard of it. You in stealth mode, I heard. Yeah, it's good that we have a user here on stage, otherwise we would have had to kick him off. So you've got a big role at OpenAI. Tell us kind of your day-to-day and actually your career path too, because you guys did cross over at Twitter and then you Instagram. How'd you get to OpenAI?
We did, yeah. I got to, I had the good fortune of actually acquiring your company, Crashlytics, when we were at Twitter. So I was at Twitter for about seven years, '09 to '16, originally as an engineer and then eventually as head of product. And then went to Instagram as head of product, was one of the co-creators of Libra, the cryptocurrency. Anybody remember Libra? Yeah, all right. And then was president at Planet, where we built satellites and imaged the whole world every day. So I've now been at OpenAI for about nine months. It's been an absolute blast. I mean, the most incredible, intense, inspiring, fun, very little sleep company that I've ever worked for. But we're having a blast and you know, it's getting to come to places like this, talk to developers, hear all the different ways that people are using our platform is just awesome.
This is all about the human experience too, so we'll talk about that. But while we're talking about sort of Silicon Valley in general, I love your take on how this current gen AI cycle compares to other cycles you guys have lived through and other moments in tech. Jeff, when you think about the opportunity here working in this space, what has this meant and how does it feel different?
Yeah, it's so much fun to build companies riding these technology waves. And to me, this is back to almost the personal computer, even more so than mobile, because mobile was a form factor. It did change how you use it, but a lot of the fundamental tech was similar. Versus the personal computer introduced like whole new capabilities, and that's where we are today. What I love about AI is it is real tech and you can go as deep as you want and really understand the magic and start thinking of how to apply it to real problems. And previous waves, I might call out crypto, I think we're lacking in some use cases. And this is just incredible how broadly you can apply it.
Yeah, I was telling someone earlier today, you know, when we were at Twitter, you're at Instagram, building products, you're starting with a, you want to understand who you're building for, you want to understand the problems you're trying to solve. And but then when you were at Twitter, Instagram, basically everything before this, at least in my career, the technology that you're building with is like kind of constant, right? I mean, a database is a database, it gets like a little bit better every year, but fundamentally you kind of know what you're working with. Whereas in AI, it's just not true. I mean, every like two months there's some new model that can do something that computers have never been able to do before and suddenly you kind of need to rethink your roadmap. You know, the problems are the same, you can still solve like really important problems, but the way you do it, it may be totally different from month to month. And that for me has made building products at OpenAI just totally different than building products anywhere else.
Let's talk about productivity because that is a big part of the sort of work life, human experience that people are already getting more productive. But from a macro standpoint, I mean, how productive can one person get? Are we going to get to the point where you are working a couple hours a day? Like, give us a little bit of a flavor of how you think about this and how much more productive can the US workforce be?
Well, I don't know about you, I spend, I don't know, a couple hours a day doing random things that like I need to get done, but the act of doing those things provides me very little value. You know, I'm like filling out forms for my kids' soccer team, random stuff that you have to do at work. I would love to outsource that to an AI. And I actually don't think it's even just about making individuals more productive. There's like, there's all of that too and that's really important. And the more you have this shift to agents, which I think is going to define 2025, right? Take ChatGPT, this is the year it goes from answering questions for you to doing things for you in the real world. You can just have it go off and, you know, do a bunch of work. Like, then it's not just how efficient can you make yourself, it's also, you know, how many agents can you have off like solving problems for you. But I will say personally, I'm not of the opinion that we're all going to like retire to a life of poetry and UBI and stuff like that. I think people strive, like it's a very human thing to want to solve problems that are bigger than you and like strive and try and do hard things. So I don't think, like, people are still going to be people and we're going to try and create awesome stuff for the rest of the world to use. We're just going to have superpowers as we do it.
Jeff, I would think accountants love this.
Yeah, I have very concrete data to share actually. So the accountant world is desperate for this technology because CPA enrollment is down about a third over the past eight years. The industry is facing this huge talent crisis. And if you look at how accounting works in America today, the typical outsource accountant works with about 10 clients, so 10 small businesses, right? And they're spending hours every month doing the finances for these businesses. We have hard data now that with Digits you can close a typical client's books in under an hour a month. And so imagine a world where now an external accountant is serving 100 clients a month instead of 10. It's literally an order of magnitude improvement.
Do we need fewer workers? We're in this new administration where immigration is a huge topic. The workforce is changing. I just wonder if you think going forward the US might need fewer workers to be competitive at a global level.
I don't know. I think again, it's human nature to want to create and to build something and to build great things for other people. I think the world is going to, like, jobs are going to change and that will happen over some period of time. But I think it's just going to be an opportunity for people to do more than they're doing today. Because I don't know, at least I couldn't imagine a world of like sitting around. That's what Masan described like 10 years ago, it was something about like a banana, like sitting back and having grapes fed to you. It was like exactly what you're describing.
You think people still want to work. So there's an economic...
Yeah, there's an economic principle called the lump of labor fallacy. This came out in the late 1800s. People were terrified that electricity would eliminate jobs, right? Every single technical evolution has had this concern. Then it was automobiles, then it was computers, then it was the internet. And it turns out there is no lump of labor that you need fewer people to do, the type of work just changes and people become more efficient and take on more work.
But does this feel different? Like, I feel like this technology is just mind-blowing in some cases when you use it and it's basically think for you. It feels like it's just such a different comparison than electricity or cars.
Yeah, but isn't there, I mean, I don't remember the name of the thing where, you know, you have a really busy freeway and you add two lanes, yeah, and then all you have is a busier freeway. Because you make it easier to do something and like more people want to do it.
Yeah, I think you're going to get a lot of that same effect. It's going to become easier and easier to do a bunch of the things in the world and so they're going to become, like, more people are going to be able to do them, which is great. And then we're all going to like move on to work on higher leverage things and like the world's going to be a better place.
It's interesting too, it feels like it's lowered the bar for engineers, for coders, or people to build a business that don't have the technical expertise. I would think that's probably another opportunity in terms of creativity and building new companies. What are you guys seeing just in your ecosystem of some of the companies popping up around this that maybe couldn't have five or 10 years ago?
I mean, I am so excited for this because there can be so much custom software that it's been impossible to build, right? Like, imagine you want an app to do X, and I think there will be a world where that all of a sudden starts to become possible and you can have a really customized tool to do exactly what you want. I actually tried this a few days ago. I was like, I really wish there was an app for this. I have three hours, can I make it happen? And the coding automation is not quite where it needs to be still, we have a ways to go, but I could imagine that happening. And so I think actually the next generation of software, we've only seen the start of it. It's going to be crazy how much software there's going to be once you can make something very easily.
You should work on your prompting.
I was upset with its technical architecture. So maybe... I was actually talking to somebody maybe six months ago who was like a social worker of some sort. And they were talking about during COVID, when in the middle of like, when none of us knew anything about what was actually going down and we were all trying to get data, wastewater, this, etc. And they had accumulated a bunch of data but they had no ability to share it. They needed somebody to go build a basic website but that wasn't their background and it was COVID so everything was crazy. And they basically, like, they were like, I could have helped thousands of people in my local district if I'd been able to disseminate this data. Wow. And I just couldn't create a website for it. And those days are gone.
That was so recent too. I mean, that right, yeah, 2020, 2021.
Like, those days are gone. Now you would just, I mean, you could do that in 30 minutes.
How does that change the startup ecosystem? I'm sure there's a lot of venture folks in the room. I would think, I don't know if the bar is so much lower to start a company, how do you assess what's valuable?
So I think the big open question is what is a defensive moat? Because you're right, if there's a hundred companies that do X, it's going to be a tightly competitive space. There's not going to be a lot of economic value created for the investors in one of them. And so I do think to touch on what you started with, AI is an incredible technology but it is still just a technology. And I think there's a lot of founders right now who are getting lost in that and aren't realizing you still need to build the rest of the product. Like, it still needs to have a great workflow, it needs to have great security and user authentication and emails and notifications and like make sure you're solving the whole problem and that problem is aided by AI, not that you're doing an AI thing and then you sort of forget about the rest of the product.
You brought up the moat. Why, how do you both think about your moat in this environment? Jeff, we'll start with you and then Kevin, it'd be interesting to hear as well about OpenAI.
Yeah, so this is what we've invested in building. Digits took over five years and it turns out it takes over five years to build an accounting ledger with all the features you need. No one is going to use a half-baked accounting product, right? There's just no world in which that's the case. And so we spent most of the time building out the accounting functionality. Like, we need to do full reporting and P&Ls and dashboards and components and it needs to be secure and you need to be able to share this data with, you're the head of marketing and they don't want to see payroll data and all of that is still classical systems engineering. And then you do the AI and you automate a bunch of the workflows.
