Futureproof by Xano
Futureproof by Xano is a podcast for technical builders, entrepreneurs, and engineering leaders who want to stay ahead of what’s next.
Hosted by Xano’s CEO & Co-Founder Prakash Chandran, each episode features conversations with innovators and industry experts who are shaping the future of technology, business, and product development.
Futureproof by Xano
Building the Next Interface — Tony Casparro, OpenAI
Have the basic requirements for building a good user experience actually changed?
In this episode of Futureproof, Xano CEO Prakash Chandran sits down with Tony Casparro, senior staff software engineer at OpenAI, to talk about his current work on ChatGPT. Tony shares some of the past user experience lessons he learned at Netflix, and how they can be translated into an AI world. Together, they cover how to get the most out of AI (and what not to do with it), some of ChatGPT’s critical UX features (deep thinking, pivot points, third-party app support), and the way in which AI is changing how we interact with software.
Topics covered include:
- From streaming to chat: What Netflix taught Tony about usability, accessibility, and anticipating human behavior.
- AI as a design partner, not a design maker: The best results come from working with AI, not outsourcing thought to it.
- The rise of contextual interfaces: ChatGPT’s new third-party apps are changing how users interact with brands and information.
- Cognitive offloading with care: How to delegate tasks to AI while keeping your creative and critical skills sharp.
Subscribe to Futureproof wherever you get your podcasts.
From Xano - The fastest way to create a production-ready backend for any app or agent. Xano unifies AI speed, code control, and visual clarity, so you never trade reliability for velocity. Sign up for free today.
You know, thinking of AI not as a replacement for yourself or your own thoughts, but as a partner to work with you is a really, really important distinction. Um, we are working together on this document. Uh, it is not just gonna have it write it and then go off uh like uh uh some people do. Um we're gonna work together on this, we're gonna solve this problem together. Uh it's a much better way of thinking through that.
Prakash Chandran:Hi there, I'm Prakash, CEO of Xano.com, and this is Future Proof. In this episode, I'm joined by Tony Casparo, Senior Staff Software Engineer at OpenAI, where he's helping shape the future of Chat GPT. Tony has spent his career at the intersection of scale and design, leading UI and platform engineering at Netflix, founding multiple startups, and now building on the front lines of AI. Um, Tony, really great to have you here. Thank you so much for taking the time. Also a fellow uh slow neighbor. Uh, it's great to have you also in town. Thanks for so much for agreeing to do this conversation. I'm happy to be here. Let's talk. Awesome. I love it. Well, why don't we start with some background? Maybe share with us a little bit uh about uh, you know, what you've done in the past and what you're doing today.
Tony Casparro:Yeah, I've kind of had an interesting history as far as working on the back end first. So doing a lot of database management, starting at Cisco Systems, uh, moving over to biotech at Amgen. Oh, excuse me, but uh really kicking off my career when I moved to Box to do um enterprise software, right? Really thinking through what are the needs of you know, B2B customers, how do we make you know great software that's easy to use, good for management observability. Um, and then doing a wild swing to Netflix. So after enterprise for a few years, I really wanted to get kind of get back in the consumer space. And that journey was really about um how do we teach the world what streaming is? This is, of course, you know, 12 years ago now. So, what is streaming? How does it work? How do you even think about having an entire library at your disposal and figuring out what to watch next? So after 10 years at Netflix, just building Netflix.com, I'm doing infrastructure work there as well. Uh, I am now here at uh OpenAI. I work on Chat GPT. Um right now we're building out uh commerce and we're also doing uh new work with, excuse me, bringing like real to like um rich responses so that the information is not just a wall of text, but actually has meaningful layout, uh responsiveness to what you're actually looking for. And so that's what I work on every day.
Prakash Chandran:Awesome. So exciting. And I definitely want to dive into all that. But I first maybe wanted to start with your time at Netflix. Um, kind of what a uh pivotal and interesting time uh when you joined Netflix, kind of in this world where people didn't realize what streaming was. Talk to us about some of like the um kind of what what the paradigm was like at the time in terms of like the concept of what it meant to stream something and some of the affordances and things uh that you had to think through at Netflix in order to like usher in this new uh era of streaming on the internet.
Tony Casparro:Yeah, you know, um I built my own servers before this. And so it was interesting, you know, that I had been used to this kind of new world for a while, but like showing other people who were, you know, still popping DVDs into the mailbox and then into their players, um, or a little bit using, you know, Apple iTunes at the time to purchase things one-off. Um, this was a new world, you know, the same with Spotify when they kind of introduced people to the idea that you have access to everything. And you're like, okay, where's my music? Well, your music is all the music. And there's this concept of like, okay, but I how do I actually use this? Well, how do I actually you know enable this to work for me? Um, so same thing with with uh streaming, right? Um, we first of all had to show people that like um here is what this movie is about, right? And doing a lot of experimentation on like showing which kind of images for the particular users and custom tailoring it to their needs. Um, do we show them a wall of text about what the movie is about, or do we show them just a few quick words as far as this is exciting and thrilling and whatnot? Um, using um instead of static assets, like actually moving assets, showing trailers in line and things like that, um, and overcoming a lot of um assumptions people had about how this was going to work. I remember going to a lot of uh tests where we would put the product in front of people and we'd be watching it behind a two-way mirror, and they'd be afraid to click on something. Why are you afraid? What's going on? Why are you afraid to click on this? Because it's going to charge me. And there was this idea that, you know, the world was piecemeal. You had to pay per view on everything you did. And the idea that it was all available. You could just click on it and it would start playing with one single cost per month. We had to educate and overcome a lot of barriers that we take for granted. So it was a lot of interesting learnings. We did a lot of experiments that worked and a lot that did not work. Some things where we assumed too much about the user's journey as to why were they at Netflix.com? Was it to watch? Was it to actually learn about a movie? Was it to figure out something to watch later and really understand uh and make a great journey from no matter what people were doing on the site?
Prakash Chandran:So, you know, I think there's so many kind of parallels in in terms of like what you're doing uh at OpenAI with Chat GPT. Uh before we touch on that, I would love to understand if there's one key learning that you still take with you today from your days at Netflix, what would that be?
Tony Casparro:It's too, uh I think it's very naive, but we we often think like, well, everyone does this the same way that I do. Uh and it's simply not true. Uh you think of something as simple as um scrolling a page. Uh it turns out there's many different ways to scroll a page. Some people click go to the right and they click on the scroll bar and they drag it down. Some people use the arrow keys, some people are using voice commands. There's millions of different, we're not millions, there's a lot of different permutations for how things people do things. And when you build an interface and you're assuming everyone thinks the way I do about how to navigate through this or what my intentions are, you're going to miss a lot of people. And so this is something that companies that you know have a lot of exposure do really well is they know, here's how to make it super accessible. A great example of this, uh, somebody else told me, is a steering wheel. Imagine if a steering wheel was always set and you had to grip it a certain way. It wasn't a wheel, it was like two handles. It only works for the way that one person thinks. But steering wheels, in by their design, allow for people that drive with it on the top of it, on the bottom of it, on the side. It works for all the different use cases. And something as simple as a wheel, it can be a beautiful illustration of how it works for everyone.
Prakash Chandran:That's really a fascinating correlation. And um, as I kind of alluded to, like, really kind of poignant to how or the work that you are doing at OpenAI, especially with ChatGPT, because it's a similar parallel because you're also designing in this new paradigm shift where people have access to everything, right? Maybe not a media library, but now the world's kind of information at their fingertips in a way that had never been accessible before, like getting to the answer versus searching through the sea of blue links. And there's an input box now that serves many different types of people. Like I use it, my uh 70 uh or sorry, 82-year-old dad uses it, uh, my sister uses it, um, my daughter uses it. It's crazy that you have so many different people interrogating and asking uh, you know, ChatGPT to do things for them. So when you think about the the learnings that you've had at Netflix and that same paradigm, how have you guys tried to approach that not problem, but I guess opportunity of serving so much of the world through this interface uh called Chat GPT?
Tony Casparro:It's a really hard problem, right? As you said, it's a super general processor. It can do all kinds of different tasks and you really have to kind of figure out how to meet people where they're at. What do they need help with? Is it relationship advice? Is it building an essay for something? Is it researching something on the web and trying to make it meaningful and accessible to them? Um, we've made great strides. Um, as were before, there was a lot of buttons on the bottom of the toolbar. We've made great strides towards you just type whatever you want, and the model is going to figure out which engine do I need to use for this. Do I need to use a thinking model? Do I need to use a web searching real-time model, whatnot, and allowing the uh the brain of the model to choose which direction will take you. There are, of course, you know, hints that you can give it if you need to say, like, this is a deep learning task I'd like you to do. Um, but overall, we were relying on that model for that contextualization. Um, and then in addition with the response that actually comes back, giving you meaningful pivot points, right? So that if you are looking for real-time information, you can pivot to like, okay, great, now show me the stock ticker. Um, or if you're looking for uh writing a document, okay, great, open this in a canvas, you know, help let's figure this out together and work on this. Um, and so it really is trying to figure out as soon as possible what is the intent of the user? What are they trying to do? And then with the response, be a really meaningful. Don't just show them a wall of text, showing them really contextual information about how they want to perceive this and give them those pivot points to then take the conversation to the next level. Um, we've got some great stuff coming up for this that really show off these pivot points. Uh, but this is key to saying, uh, to showing or with hints, showing the user the capability of the product without being overbearing and showing them a wall of text or buttons.
Prakash Chandran:Yeah, you know, I think um even in coordinating this, um, we had to make sure that you got through uh a crazy amount of work, uh, obviously culminating in just the recent developer day that happened. It was kind of an amazing moment, actually, for me to watch because I really feel like ChatGPT has evolved in uh in something that was maybe considered like a chat interface and tool to more of a platform and to what you're speaking about, giving people more tailored contextual experiences rather than just a wall of text is something that we can just see the world uh going to. Tell me, like, as someone that works on the interface, how what are some of the key challenges that people not may not necessarily understand or think through that you have to think about in terms of presenting that type of information with context and tailored and personalized to the person that's searching?
Tony Casparro:What's interesting is especially in this world where the model is trained on a lot of information and yet there's a lot of real-time information that it can have access to, real-time information in terms of what's available on the web or tools out there, all these new um uh MCP servers and whatnot. Uh, and so one of the complexities is how do you orchestrate giving the user a fast response and at the same time, basically in parallel, let me go talk to the stock widget, let me go talk to the weather widget, let me go talk to this MCP server and bring this back. And so some of the complexities of the UI are UI are you know, how do you show the information that you have so far? How do you show the user that you're still bringing in extra information or that you're processing something really important and make it not feel slow and janky, right? Because the power is that it can build you a great response. But if it feels like, man, nothing's happening, I don't even know why I'm waiting anymore, that feels bad. And so what we do is something called uh the chain of thought. So the chain of thought is as it's thinking through whether that is real-time information or talking to other servers, we're showing you little snippets. Hey, I'm talking to that server over there right now. I'm gathering the real-time weather information. And that's giving the user insight into what's happening behind the scenes in terms of a thinking model that's interacting with the world, um, and then building that useful uh response in turn.
Prakash Chandran:Yeah, that orchestration and kind of giving a sense to the user around what uh ChatGPT is doing is just it when it works really well, it just feels like really seamless and elegant, but I can imagine how much thought goes behind it. Um, you know, one of the things that we've talked about before uh is kind of the most uh underutilized uh tool within ChatGPT, and we've talked about it being deep researcher. That that was the answer at the time when we were chatting. Uh and since you told me that, I actually have been using it a lot more. Even though, like, you know, I think I'll I'll be in a conversation and uh it will obviously try to optimize for quick answers, but I'll explicitly say, look, for this one, I want you to think deeply about it. I'll make sure to tell it to go into deep research, and it gives me like something really, really thorough. I'm almost giving it the freedom to take a little bit more time to give me like a really thorough answer. Do you see more people starting to use ChatGPT this way? Because I think this is what I'm imagining. You tell me if this is true. People, a lot of people, are shifting their behavior from traditional search to Chat GPT, but they're still using ChatGPT like search, right? They're still typing in like maybe piecemeal words and not necessarily using it how it is meant to be and how it can best be utilized. Tell us what you're seeing and uh kind of the dynamics around like deep research as a feature and the best way to utilize ChatGPT.
Tony Casparro:Yeah, you know, um, so the deep research is one of those hints that I was talking about. It's in the toolbar where you can say, This is how I want you to answer this question. And deep research specifically means take as long as you want. I don't want a quick answer. Um, I want you to be very thorough, go through lots of sources, take your time, and get back to me with a really deep, full response. And so it is one of my favorite features, especially for shopping for something and comparing, you know, a bunch of different review sites or models uh for travel advice. Um, it can be really, really thoughtful instead of giving you an answer very quickly. Um, when you ask about user behavior, though, it turns out people still want something very, very quickly. They don't want to wait, they want the answer right away. Uh, and so what we've kind of come up with in terms of the UI is kind of a bifurcated mode. So it will start to say if it is going to take a while for something like I do need to do some research, whatnot, there is a skip button. You can say, I don't care, just show me what you got right now. And that allows people the flexibility to be like, uh, this is how I uh want my answer. Um and so I can't say for sure like what the adoption is like on some of our uh other features like deep deep research, uh, but people are still very much in the mindset that they want it quick, they want it fast, and they want it correct. And we are working as hard as we can to make sure that that is the path forward.
Prakash Chandran:Yeah, I love that bifurcated mode where it's like, hey, look, and I think it says something like thinking deeper for a better answer or something. And then you have the ability to be like, nope, nope, it's all good, just get give it to me now. Um, and I think that's the hard piece, right? Kind of balancing that speed, that AI speed that people have come to expect with the thoroughness that something like a deep research um can offer. Um, I want to shift gears a little bit. You know, you are a staff engineer, you have been developing in some capacity for your entire career. Uh I think the audience might be curious to know how you are leveraging AI in your day-to-day to develop software at OpenAI and uh so you know some of the core benefits that you've seen and things that you might have to like uh supervise or kind of still be aware of that are still kind of um evolving in terms of like the landscape.
Tony Casparro:Yeah, absolutely. Uh, you know, the best way to think of AI is as an assistant, right? It is there to help you with your tasks. And so when I get into the office in the morning, what do I do is I look at my plate and I go, what can I offload to an AI so that I can work in parallel? So some requests might come in for some code changes or connector changes or whatnot. Um, I can muddle through those and go do everything in a serial manner, but I have the ability now to say, let me offload some of these tasks. So we have Codex as one of our um AI coding agents. It's available to everyone as well. Um and it's pretty phenomenal when it has access to your code base to say, you know, this is the change that it can even read, of course, now like linear tasks and Slack messages. And you can say, here's the context of the issue, come up with a solution for it. Now, uh, like all AA things, it may not be perfect at start, but it can actually give you quite a bit of leverage and thinking through opportunities so that when you come back to the task, you can be like, well, that's pretty clever, but let's change this, rewire this, perfect. Now we're done. And so as opposed to me working, like I said, serially in the morning, I'm thinking uh across many different verticals at the same time. I've also got AI, of course, built into my IDE, which is my uh coding interface. And now it's like, you know, like, you know, applying some formatting or markdown or just kind of threading through some variables and whatnot. And I can say, just finish this off for me. And so that allows me to have a much more productive day in terms of focusing on the bigger picture tasks uh and then uh sharing the work with the AI to take care of the more kind of menial stuff to like just kind of thread through. Um, and so that's for work and then also for you know doing research and things like that for other things that we have going on at work. Uh it used to be a world like where you said, like, okay, I'm looking for this new um state management. Let me go to Google. Okay, I see a bunch of links. Let me go read through these articles and spend half my day doing that. Or I can say, hey, here's another task for you, not Codex, but ChatGPT. Go research the latest state management to make sure the information is up to date. Bring me a report, build me a spreadsheet that I can review. And that is once again, not doing all the work for me. It's in a supervised way where it's helping me and assisting me, giving me a collection of information that I can then use to make my next decisions and go deeper as I need to. Yeah, that's fascinating.
Prakash Chandran:So, like there's gonna be developers and technical builders listening to this. And one of the things that you said was kind of really um uh really interesting, just in terms of like, you know, at the beginning of your day, you're kind of figuring out how can you leverage this assistant to take care of some things, to do some research for you while you focus on maybe the higher order bit. For the people that are listening, how do you structure that, that planning process around what you can delegate to the assistant and uh to uh leveraging AI for versus what you need your attention on?
Tony Casparro:Yeah, I mean, you know, so as a UI engineer, uh I do a lot of visual testing. Um, and it's something where uh our like Codex is extremely good at like backend coding right now, and it's still developing its chops in terms of UI coding and testing those UI changes. How do they work? How do they work for accessibility? You know, how does it work on a fast connection, solo collection? There's a lot of things we need to consider when we're building user interfaces. And so some of those tasks, I might have it like um, for example, you start building this for me. For example, I was building an animation the other day, and I knew it was not gonna hit the animation correctly, but I was like, there's gonna be an animation that occurs when somebody clicks this button. I need you to pop open this, do this, use this variable. And it did a pretty good job, but at least it was a start for me to be like a jumping off point for my PR that I was gonna eventually write. Um, and so when I come in the morning and I'm thinking about those tasks, uh, I am thinking uh in terms of like, how would I approach this problem? Okay, these are the steps I would take. Let me give those to the to the agent and have the agent take care of it for me. If that is too complex, I go, I can't even fathom my head what this is gonna look like. I just have no idea what the outcome is going to be. I don't know uh the correct way to do the performance of this, whatnot. I'll take those on and start on that journey because it's it's just an unknown. So to clarify for short, if it's a known end state that I want, uh once again, comparing things uh across different websites, doing research and analysis, whatnot, there's a known outcome. I want to find the best this. Uh, that's a great choice for it. But when you're really ideating, you're not sure what it's gonna look like, you're maybe working with partners and stakeholders, and you're like, we don't know what that's gonna look like yet. That's more the state where I take ownership and control to like do the human stuff, talk to people, get feedback, and figure that out too.
Prakash Chandran:That makes sense. Um, you know, talk to me a little bit about where you feel like uh AI is not quite there yet, but you're excited to see it evolve so you can take advantage uh of leveraging it kind of in your day-to-day.
Tony Casparro:Yeah, we've made great strides towards um orchestrating tasks, right? So this is the idea, like in you know, a year ago, if you said, you know, change this piece of code and structure the state in this new way, it would have just started spitting out code, right? It wouldn't really think through its tasks. And so something we've told these coding agents now is develop a set of steps that you're going to take, right? I'm gonna do this, think through these, think through the uh side cases and edge considerations here, and then start coding it. And you get much better results when you have it think through a process to follow first. Um, and so um, in terms of not being quite there, the UI piece is is tricky, right? Because you need to test across different browsers, you need to test across different connections. Um also AIs don't really know what a beautiful animation looks like. They don't know how it feels to a user. That's something very visceral where you go like it should feel this quick and move this, and you know, especially when you're doing multiple uh coordinating multiple animations in the same way, you need that to feel really good together. And so it doesn't have a good feel for that just yet. Even in terms of the user interface, I've experimented with a lot of um our own tools and competitors in terms of like what makes a great UI. And they guess and they look okay, but it's kind of using just random examples. It's I wouldn't say it's a great UI designer yet. Um, there's something really magical about the way humans think through how a user interface is going to flow and how that's going to work. And so I think those are areas for opportunity. Backend stuff is getting very, very good at. Um, you know, if we go to a world in the future where um backend code is not even a code box anymore, it's just a black a black box where you say, hey, uh, this is the inputs of the API I want, and these are the outputs I expect. I don't care what the code looks like inside, uh, as long as it's quick, just go be a black box. And we may end up in a world there where we have black box, you know, um inputs and outputs for Vatican servers.
Prakash Chandran:Um, you know, talk to me a little bit about how uh just in a from a development standpoint, you think about like security and control over what can sometimes be the back uh the black box, uh, whether it's front-end or back end, how you think about that in your day-to-day? Because I feel like that's probably the supervision part that you're talking about, making sure that you're in the loop and making sure that it's doing the right things. But you know, in your day-to-day, how are you kind of implementing that and thinking about it as an engineer?
Tony Casparro:Are you thinking, are you asking specifically like how I use the security controls or how we use the other thing?
Prakash Chandran:How is you you as a company think about those controls in the develop in your day-to-day development?
Tony Casparro:Absolutely. So it's person in the loop, right? You know, uh the AI should never have access to do um what could be considered um uh dangerous tasks on its own, right? You know, executing you know, arbitrary code and things like that. It will ask you and prompt you, is it okay if I run this? Here is the exact command I'm going to run, and make sure that the human is always in control of those tasks. Um, as we start to get an exposure into the world of MCPs, same thing. We want to make sure that those uh checkpoints are there so there's human in the loop and so that it's not off and running on its own. As our developers, however, build their own products, we're encouraging them to also put those same safeguards in place. However, there might be safely guarded loops where you go, like, hey, it's gonna generate images. I already have these moderation controls in place, I know this is going to be safe, and you can have it run that uh with that oversight, uh, as long as uh a developer or somebody has already made sure that there's a safe path for it to take.
Prakash Chandran:So I want to go into just maybe some practical takeaways for people that are um, you know, might be building a new project and trying to figure out the best way to leverage AI. You know, you feel like there is, you've got all these bypoders and they're you know building applications very quickly. You have people that are leveraging AI alongside of themselves in a code base, uh, but from a chat GPT perspective, for a new builder that is trying to think of uh, you know, developing something new into the world, or even for a product owner in a company that's trying to make something new within their organization, just in terms of a mental model of how they should be thinking about it and leveraging AI, what recommendations might you have in order to get started in the most efficient way possible?
Tony Casparro:You know, it is interesting because we have Codex internally, it has access to our Intari code base. And so we've had people, uh, I mean, that's always a great starting point. If you have an existing product, you give it access to your GitHub repo, it knows everything about your code. And when you ask it a question about it, um, it can efficiently use the current styling and permission sets and uh plans that you use for your existing code, and then it can do the augmentation or adding new features, whatnot. That turns out to be decently well. Uh, we've had people that are product managers, they asked for those changes, they send it to like a developer with the change set. Uh, the developer can then do slight modifications or whatnot is needed, but it does help the process along because um usually that process is the developer going to the person, the person describes what they want, they go back and forth, and it's nice if you know the person can just describe everything you want up front and get that result. Um, in terms of like starting new code bases, uh, that is a tricky one. You know, zero to one with a new product. You know, um, I personally wanted to do like, you know, some more app development and whatnot. Um zero to one is always a hard proposition, right? Because you have to build out some servers or some uh, you know, some developer toolkits on your local machine. Um I can't speak to any websites that currently do like a great job. I'm sure they're out there, uh, but it's a great opportunity because there are so many people that are vibe coding small applications. Uh, it'd be fantastic if you could uh use one of these to start building up, you know, whether that's a website or an app that you're gonna release to the app store uh and get you up and running quickly. So uh I guess do some research. Uh I'm not I can't recommend any particular ones. Uh Codex is great though for existing codes.
Prakash Chandran:That's awesome. Um and then, you know, just in terms of like uh, you know, ChatGPT is such a challenge for all of the reasons that we've been uh speaking about, just in terms of the U and I UI UX uh paradigms that need to be introduced. I'm curious to uh if you have thoughts around any UI or UX patterns that you will see either disappear or be introduced into this new world of new interfaces uh that we're entering into.
Tony Casparro:So that's a great question. That's right in my wheelhouse. So um uh as your users may not, or your listeners may or may not know, uh we released something called uh third-party apps inside of ChatGPT uh just this Monday. Um so this is the world now where you can talk to ChatGPT, and instead of ChatGPT saying, here is a link to Zillow, go like click out to them and leave the site. Uh you can now talk to Zillow or these other uh apps that we partnered with to start inside of Chat GPT. So um that means they're creating custom UI elements. Uh ChatGPT and these uh third parties are now working together to build visualizations. Um and that turns out to be extremely powerful. Another example is Spotify, right? Um hey, I'm coming up uh for a playlist, we're having an 80s party, make me a great playlist. It could list a bunch of songs, but it'd be great if it could interface with your Spotify account, build you the playlist, and then you're done, uh, as opposed to you having to go add those songs manually or whatnot. And so bringing these apps together to be controlled inside of ChatGPT is gonna be a really interesting new world. Excuse me. And then, of course, on the other side of it, we have these agents that you can build into your current sites that go backwards, right? So now you are the main app. Chat GPT is now embedded inside of your application. Now these two can interface as well. Um, it's gonna be a really interesting world in terms of how people think about you know, branding and um app stores. I mean, you can see a world now where if these UIs are so like chat agents are so powerful enough to build, like integrate with third parties and even build interfaces on the fly, you could start to see app stores go away. You know, why would I have to go like search my phone and find this one app if this other app can just do it all for me, right? It can generate the UI, generate my feed, whatever I need, and build it in a custom way. And so you can see a long tail road down here where, you know, uh there is no more searching for apps, there's no more looking for context. You can just interface with your device in whatever way you want, probably through um speaking and or text, and just say, show me the weather for today. It doesn't have to open an app, it doesn't have to do anything else, it can render a rather a weather widget for you. Uh, you know, tell me how my finances are doing, you know, like cross-reference my portfolio and see what I need to do for my taxes. And it's generating stuff on the fly. Uh, and so there's a really powerful new world out there for those that um are thinking through how does AI and my application fit together to create a single united world as opposed to I gotta get back to my site. Um, and it's gonna be real interesting to see.
Prakash Chandran:I totally agree. I was actually talking to a previous guest um about this, how like the front end uh, you know, in terms of the way we're used to digesting, where you I download a singular application or I go to a web application, uh, is changing. Like because uh the more people interface with tools like a ChatGPT and get those answers, yeah, the nature of how companies will express their brand through like this new apps, SDK and otherwise, is going to just fundamentally change. So we're kind of entering this era of like personalization. And uh and another thing that he he brought up that I thought was fascinating was a potential world of like ephemeral uh applications. Because sometimes we build an app for a purpose, but once that purpose is done, especially with how fast we can create applications, like it may have be like purpose built for this one thing and then it may go away. I'm curious, have you have you thought through anything like that, just even outside of what uh you're doing at OpenAI, but just personally, how do you think about the world of um digesting information, application development, and how we're going to utilize or interface with businesses in the future?
Tony Casparro:Yeah, you know, it's funny. Netflix did a lot of kind of one-off applications that were short-lived. You know, we were doing experimentations inside for, you know, do, you know, does it matter if we show a different image to a customer for a movie than this other image, right? And we would do these experiments where we would have an idea like that and we would need a UI that was just, you know, ephemeral. It was just gonna be there for a while while we evaluated the efficacy of this product. Um, and we would spin those up all the time. Um, and that process was you know very cumbersome at the time, and now we have AI to help generate some of those for us, and it's very, very nice. Um, in terms of like, you know, um at ChatGPT as well, it is amazing because when you think of like what chat can do now, um, you know, it more so, like, especially with my team, more than just a wall of text, right? We can do tables, we can do interactive tables, images, videos, and things like that. Um, you're getting to a world where, like, yeah, uh, I can create, you know, I'm I'm I'm learning something, you know. Let's say I'm a student, right? I'm having a little trouble with understanding acute and obtuse angles. Uh, you know, build me a quick, you know, uh interface, uh, a program that I can learn this from. And it goes, boom, it just generated for you within seconds. And you go, Oh, I see how the angles work. Now it's interactive and it shows you, and it was custom to you. And how that is better than kind of what we have today is, you know, today somebody may have already generated that tool and you have to go to the website and to find it, but this is built inside a chat. You already has the conversation and context, and it probably even knows that you like Minecraft as well. And it made the tool fun and interactive, like because you know Minecraft. And it's building and creating these ephemeral tools that are unique to you, to your learning style. I think it's going to be such a huge boon for uh, you know, children learning in schools or also even, you know, adults that are learning new concepts every day, especially in this tech world. Like, okay, I know about this, teach me about that, and it can do it in a way that's really relevant to you. And so um, I'm excited about those quick and new interfaces and tools that it would be able to generate. Uh, we're not quite there yet.
Prakash Chandran:Yeah, for sure. I cannot wait for that. Um, you kind of touched on something that I have been thinking about a lot, just even with my own personal usage, um, you know, and that's literacy, right? Like I think uh the amazing thing about something like a Chat GPT is I'm able to deeply kind of think through with or sharpen my thoughts around something by just interfacing with it. Like I'll be driving and kind of using the voice feature and having a conversation, and I don't feel embarrassed to be like, can you explain that again? Like I'm like eight years old, please. Um, but it really, really has helped me clarify um a lot. On the other side of things, I've found myself like, you know, ChatGBT is so good at, you know, generating, for example, email responses or even things around structuring things for my board deck that I'm like, I don't want to sacrifice kind of that skill that I have of doing that myself. I feel like there's like, there's like trying to give it too much and do too much for you and losing kind of that ability to shape it yourself. And then also kind of leveraging it to like help sharpen your thoughts and like from my my personal point of view, using it appropriately. Like that's like the highest and best use for it. I'm curious how you personally think about that, right? That spectrum of like handle it for me versus using it in a way that's productive and doesn't take away from your uniqueness and the value that you provide to the world.
Tony Casparro:It's an excellent question, um, especially as we're we're using it more and more for everyday tasks. Um, somebody has coined this as cognitive offloading, where something that was challenging for you, that's you know, stressed an ability that you have, you're now offloading to the AI. Um, I have caught myself doing certain sometimes doing things where I go, you know, I should be able to figure this out. You know, maybe it's a multiplication problem or whatnot. Let me do this. Um and it's important to keep in mind that uh as humans, you know, stretching our mind and keeping it sharp uh is really, really important. Using your literacy tools and using uh your abilities is really important. So, you know, thinking of AI not as a replacement for yourself or your own thoughts, but as a partner to work with you is a really, really important distinction. Um we are working together on this document. Uh, it is not just gonna have it write it and then go off uh like uh uh some people do. Um we're gonna work together on this, we're gonna solve this problem together. Uh it's a much better way of thinking through that. And so being careful uh to not tip that too much in the favor where you're um uh not stretching and you know use utilizing your brain as much as you should, and remembering that that's really important for you. Uh the using your brain and stretching it every day is uh critical to staving off a lot of diseases later on.
Prakash Chandran:Totally. And I think um I think just the kind of the key thing that you said there is making sure you're working together and learning together alongside uh of it. Because I think for me, at least that has been kind of the best use of it. And I caught myself once where I was like trying to have it like answer everything for me because like when you have this kind of magic thing, you're just like, oh my God, I'm just gonna use it for everything. And then I ended up, I remember, copying a response to like I had mostly written the message, but like I just had relied on it, ChatGPT, to kind of finish it up. But then I pasted it and it pasted also the um, okay, here's like an updated response that maintains your tone and everything in the uh the like a Slack message that I had sent. And it just it caused me to pause and be like, okay, did I really need to do that? No, probably not. But like I was, you know, it's so good that like I I had started to kind of go in a way that me personally, I wanted to kind of co-create and learn alongside with it. So I feel like I've gotten into a good balance, but we'll see as things uh a great example of this is a power tool.
Tony Casparro:I remember, you know, my dad taught me, you know, growing up how to use a saw, right? And like, you know, the pain that it was. And using power tools is a whole other ball game. But you have to pay attention. You have to really make sure that you are in control of that product. Um, otherwise it can really hurt you. And so that is the same way I approach AI, right? Like I am, it is a tool that I use, uh, but I need to make sure that I have always control over it.
Prakash Chandran:Um, as we start to close, just uh one more question here, just on something that you had mentioned, which is like, you know, we talked about kind of the age of like this new interface, this personalization and you know, the AI or ChatGPT being able to like tailor things from a number of different data sources together to give the user the exact right thing. If I'm a business today and I'm thinking about the interfaces that I'm gonna design, I'm starting an application or I'm trying to expose my business logic uh in a meaningful way, how would you recommend thinking about that? Like from the interfaces that they're creating to um, you know, their integrations with AI applications, like where would you recommend that they start to make sure that they're future-proofing themselves?
Tony Casparro:Um, you know, so if we're talking about like a company that's making things making things for consumers, uh, first of all, um, most people are starting to go to this role of using voice or text to describe what they want, right? So if they come to your site, if you're having them navigate through a lot of, you know, wizard wizards or menus and whatnot, um, people are really coming to this world to expect, like, here's what I need to do. Like, you know, give me a couple options or you know, allow me to use natural language to tell you what I want and give them that approach through their site, right? Making people learn how your tool works and the complexity of that. Um, maybe there's a place for that, but most people are expecting now for the UI to be tailored to them instead and not the other way around. Uh and when you talk to uh business, business customers now, uh really making sure that you have excellent documentation, you have the ability to search and ask questions in it. Um, developers get very, very frustrated when they can't see a lot of good examples or interactive tutorials or ask good questions of the documentation. These are kind of now stakes, uh table stakes for people coming in to work with your tool. And so ensuring that um that is really primed and ready to go. I'm sure there's third parties out there that offer this as a service to set those guides up. Um, but that is really, really key. Um and so yeah, that is kind of the way to win people over right now. People are expecting easy, they're expecting low friction. Um, and if you're like, hey, you have to read all this and figure it out and muddle through what we we've written, and we're not gonna give you any guides or help for on it, it's like, well, just making it harder, you're reducing that friction they know.
Prakash Chandran:I I heard someone else kind of talk about there's kind of the user experience, but also kind of like an agent experience where you have to make sure that things are really well documented and described, just like you would for a human that like needs to understand all the elements of it. But when you have also a machine that's working alongside that human that can digest things much, much faster, it's really critical that your business logic, the way uh the rules and the processes that you want people and machines to understand are really well documented. So that makes a lot of sense. Tony, you've obviously been working on lots of different things. There was a dev day. I'd love to hear about what you have been working on lately, what's been keeping you occupied.
Tony Casparro:Yeah, so we uh just released uh Commerce inside of ChatGPT. Uh previously, if you were looking for products like you were I'm looking for a new expresso machine or some shoes or whatnot, we would show you a list of links inside of there. Um they're really the links are not ranked in any way, they're according to what you want. Um, but there was no ability to actually check out inside of ChatGPT. So one of our first kind of like, you know, these new third-party products inside of Chat GPT is you can just hit the buy button and it appears in Chat GPT. So we work with Stripe as a partner to do all the credit card processing, and then we work with our um merchant partners, if that's like Etsy and Shopify right now, to do the transaction. So this is agentic shopping where we say, here, we'll collect the credit card information, we'll talk to the store, we'll orchestrate the transaction, it's done. Uh, we were involved in it only simply that we have a train, a copy of the record so that you can easily access it from that chat conversation, but it was all done for you between these different partners. So they facilitated transactions, they did all the um in credit card encryption, whatnot, and then the shipping. Um, and so there's this new world of once again, inside of Chat GPT, you have control over things happening outside to even do your shopping for you.
Prakash Chandran:That's amazing. So you're really just kind of a facilitation layer, and then you're like bringing these partners together to make sure that that transaction happens based on the user's intent.
Tony Casparro:Exactly. And it's great for the user. We've got the credit card information on file, nice and secure on our side, well, stripe side really. Uh, and we're simply making it easy to hit that buy button inside of there and do this, do all your shopping right there. That's awesome.
Prakash Chandran:Just as we close, you know, coming off of all the hecticness uh of developer day and everything that you're doing with OpenAI, if you had to leave the audience um with one thing, just something that you would like them to invest more time doing, whether it be chat in within ChatGPT or otherwise, what would that be?
Tony Casparro:You know, shockingly enough, it's simply give it a chance. Um, I have many friends who are in this with me. They're co-u-engineers and whatnot, and there's still a lot of apprehension about, you know, um its access or controls or if it'll actually be useful. Uh, and honestly, give it a chance. Uh, it's the same way it was with Spotify. I remember people telling me, like, it's great. You have these, these, you know, all the music at your fingertips. And I go, like, no, I I like the way I've always done things with things with CDs. I don't want these new tools, and I didn't really give it a chance. Um have it do something, you know, where you know it might take you a half hour to like search through something on the web or do some piece of research, or you're looking for, you know, everybody's always looking for, you know, a new product or, you know, an espresso machine or whatnot. Have it do some of that task for you and see see how well it can do at that task. Um, if you're willing to have it wire up to your code base and just say, here's a task I was gonna do today, do today, give it a try for me and see how it does. Um, these we've made these things trivial easy. Like you don't have to log in, you don't have to make an account, you can just go to chat gpt.com and just ask it a question and try it out. Um, it is uh it couldn't be less friction along that path. Uh, and it might meaningfully make a difference into your day-to-day life. You might be able to focus on the higher level stuff that makes you a better uh um engineer leader in your workplace. Uh, you might be able to leverage your time better, uh, and that makes you a better effective uh person in your field uh and uh all the more powerful.
Prakash Chandran:Tony, what a perfect place to end. I really appreciate your time today. Thank you so much, buddy. Thank you very much, Rakash. Appreciate it.
