Why OpenAI Is Racing to Make Computer Use Superhuman: transcript

We've gotten computer use now to a place where it's faster than the average human and I think we're getting really close where it is superhuman. >> What sorts [music] of tasks do you think are ready to hand over to the agent versus the ones like okay, it still requires some checking [music] and correcting. >> Computer using agents now since Astra is now capable of doing anything that a person could do. We think a lot about what new consumer experiences can we enable with a faster better model and then on the enterprise side that's kind of naturally why the safety becomes more interesting. How do we give the ability to control the amount of human in the loop there is and then the auditability.

>> It's unlike any other thing that we've built where the developer has to do some work to adopt it. Someone has to build a connector or MCP server. You know, computer use works with everything. >> Now we have like capable models. They're getting cheap enough to run at scale. What should companies think about automating? I think of it as in two big categories. The first category is like >> Welcome to the a16z podcast. I'm Sonal Ambani. I'm joined by my colleague Fabrizio Serafini. In August we wrote about computer using agents and how they're starting to do the real work inside companies.

Today I'm joined by Ari Weinstein and James Sun from Open AI who are going to help unpack what is going on with computer using agents from their perspective and share some thoughts on the future. Ari, James, thanks for joining us. >> First happy to be here. >> Yeah, thanks for having me guys. >> Ari, kick it off and tell us a little bit about what in Astra has made computer using agents move forward and what what's really unlocked. >> Yeah. So Astra's an amazing model. It's good at so many things.

Our computer use models were already so good before Astra. You know, like Soul itself is amazing at computer use. So Astra really helps with a few things around the edges in my sort of anecdotal experience. One is it is a little bit faster on average because it uses fewer thinking tokens to do the same task. It's better at certain kinds of creative work. Like for example, we had this demo where in our launch video where you were talking to it and it's drawing things on Microsoft Paint. It's way better at that sort of thing, which I think probably transfers to other kinds of creative tasks also. And then the last thing I find is it's just more reliable around the edges. Like, you know, it's kind of I feel like I trust it a little more.

Like, it's sort of the kind of thing where there's a little difference that is hard to perceive but really important between something being 99% reliable and 97% reliable and I feel like it's sort of just a little a little higher in a way that feels pretty pretty concrete. >> Yeah, and I think adding on to some of those things, I just also our most aligned model, yeah. Right? And so, you know, talking about trust a little bit. Yeah. You know, if you look at the system card for it, it is much better at being resistant to prompt injection. It's much better at not taking any kind of unintended actions. And so, when you think about it for like enterprise work, especially for companies using it for sensitive things, whether it's like procurement or you know, talking to vendors or talking to customers, it's going to be just you're going to feel a lot safer kind of using it and that's really really exciting. And then there's some some fun things on the creative side. I think internally we had you know, Astra play video games against you know, 5.6 and it wins every time. And so, there are some cool tweets that people have made where you know, Astra

is able to beat Fallout in like 30 hours or something like that. And so, it's also just really amazing at playing video games. So, I don't I don't know, that probably translates to something but >> [laughter] >> yeah, it's pretty exciting. >> just double click on that point. Um, consumer versus enterprise use cases of computer agents very very different implications around kind of like all the scaffolding around I would say the computer using agents. How do you guys think about where things are more exciting between consumer and enterprise if if you were to pick?

>> I think our goal at OpenAI is really about building products that's useful for everyone. And so, Um, know, actually it's interesting cuz Ari is definitely more of a consumer guy here and I spend a little bit more time on the enterprise side. Um, I think the consumer is always the place where we try to push the frontier of what's possible. Uh, you know, uh, with Astra like Ari mentioned, there is it's a much faster compared to the previous models. And so we think a lot about like what new consumer experiences can we enable with a faster, better model. And then on the enterprise side, that's kind of naturally why the safety becomes very interesting because I think that the enterprise, whenever we talk to a large companies, I think the use cases for computer use is actually fairly obvious, right?

Inside enterprises, there's lots of legacy software that doesn't have any MCP tools. They are software that's been used for like 15, 20 years. And so there's and and and a lot of times that's actually kind of hard to use for a normal person. Lots of like UI that is like from the last, you know, decade. And so you don't even know how to operate some of the key key key things. And so we've seen companies build skills that kind of teach the model to be able to use them better. And so I think maybe the way the way to think about it is enterprise is where, you know, the capabilities have been there maybe for a little bit longer and the companies are catching up and trying to understand how to deal with it. And consumer is where we're trying to figure out the next types of use cases that's not maybe possible today with software. Maybe you can share a little more.

>> Yeah, I mean, the consumer use cases are so fun for me because it's kind of like how can you put your agent to work doing the things in your daily life that are so annoying and save you tons of time and just have more time to focus on the things that um that you're excited about. Um, I think it's been really fun to see Dot uh where, you know, each Dot gets its own computer and you can ask it to do things for you in the cloud. Um, it gets a full web browser. It can buy things for you. It can, um, you know, book haircut appointments and uh you know, anything else you want it to do. So um so that that's been really fun and I think there's a a lot more on the horizon.

It's going to be really fun. >> Yeah, no, that was going to be my question on on on Dots. I think James, you probably use a dot to walk through product feedback. It'd be cool if you could share share more about it. Or I'm just so curious to know on on your end, guys, like how are you using dots and like how does com- computer use play into kind of why you can do with them? >> Yeah, I mean, I think dots is really cool for a few reasons, right? One is like really proactive and so I've been very surprised a lot of times when in both my personal life and work when dots would just flag something that I just completely slipped through the cracks, right? Just the other day. So, internally like on our our email system, whenever we get an external email for an invite, it doesn't actually show up on our calendars. And so I end up mis- missing a bunch of external meetings.

And so dots now is like catching these things for me. It's like, "Oh, there's this invite that's in your inbox. It's not on your calendar. me to like accept it in your Gmail so it shows up?" So, that's been a lifesaver. It's a small thing, but it matters a lot. Um but I also have it proactively kind of looking at I think it's really good at proactively looking for tasks where it's the shape of it is not super uh let's say concrete every single time. So, an example that I also have it set up to do is um we use Slack internally and there's there's a ton of messages that's going on at all times. And uh I used to have this Slack job uh using automations in Coda X that's like every hour or every 30 minutes go and try to pull things.

And I've just ended up tuning out a lot of the notifications because it's just like sometimes useful, sometimes it's not. With dots I can give kind of a high-level goal. It's like, "Hey, if anybody in my reporting chain says anything, flag it to me, right? If anybody who's giving some critical feedback, use your judgment uh about certain products, go for it." And then because it has a cloud computer, um you know, I've had this pretty cool workflow where when someone reports, they usually have like a Sentry log. I could have it go use the cloud computer, um you know, actually drive my local local browser while the dot's going to con- control both your local computer as well as the cloud computer, log into century, pull the logs, analyze it, give me feedback, and so it's saving that entire loop of figuring out which feedback is important, looking at it, then going to century, downloading the logs, feeding it into code X, asking for thoughts, and all that can now just kind of be automated. And so, um yeah, that's is great for for at least these proactive and kind of long-running tasks where they're not the same shape every single time.

>> We talked a little bit about reliability, speed, token usage, and efficiency. How do you guys think about balancing all these things as you as you keep making improvements on what you're what you're releasing? >> I think we're really excited about computer use at OpenAI. We've invested in it a lot, and we've gotten to a place where I think we have the fastest, cheapest, and most reliable computer use product. Um and I think there's actually so much further we can take that. Um we're we spend a lot of time looking at speed. You know, speed is really interesting because on one hand, the power of computer use is it can do tasks in the background, so in some cases it doesn't matter how long it takes because you can going and doing other things while it's accomplishing the task for you. But also, sometimes you want to watch it to make sure it does it correctly. You need to wait for it because it needs to complete this before something else can happen. And so, the speed is so important because um you know, you don't want to get this feeling like, "Oh, I could have just done that myself, and it would have been so much easier." So, um you know, we've gotten computer use now to a place where it's faster than the

average human at using a computer, and I think we're getting really close to a place where it is superhuman, meaning it is faster than an expert computer user like probably one of us at this table. Um and I think that's going to actually be a sea change when that happens. I think it's going to change the way we all interact with computers because all of a sudden, anything that you you know, that that you could do manually, it could do faster. So, unless you really want to do it yourself for some reason, which I think actually there may be many reasons people would want to do that. But in a default case, you probably just want the agent to do it for you, and I think that's going to let us build all kinds of cool user experiences. So, speed is something we think about a lot.

And then, obviously, we have a strategy of offering multiple different models at multiple different, you know, qualities and price points. And so, for different use cases, you may want to use a different model. Um at the same time, our um our less expensive models are getting much better at computer use over time. And so, I think there will actually be more variation over time in which models people want to use for which tasks. >> I mean, that's what you're saying. Like, I think the other thought is like, how do we give customers, you know, control over the safety, reliability, autonomy versus speed, right? Like, if you have like, you know, when when the model is slow, you can kind of watch it do stuff. But then, you're limited to that to the task that it can do because you can't just sit here watching the thing doing it.

Like, it's actually less effective, right? Because it it's slower than than you. Um but once you can get to a point where you can no longer observe and audit everything it's doing, um how do we give, especially enterprises and businesses, the ability to control um the amount of human the loop there is and then the auditability, right? Like, I think a few months ago, I was part of Astra just a few months ago. Part of Astra, you know, we we kind of launched a set of enterprise controls that let um businesses choose one, like, how much human the loop is there? Like, do they want the agent to automatically approve everything? We have this really awesome system called Guardian, right? Our auto I think auto approve in the app that uh is a really smart model looking at every single action that the computer use agent is taking to decide if it's safe, if it's if it's aligned to the written intent. Um a consumer might be like, I just want it to do everything, and I don't want to be asked for approval. Whereas, an enterprise is able to kind of decide which type of actions actually go through. And so, I think the the the control part becomes more and more

important as our computer's agents become also superhuman in doing things. >> Totally. I think as you try to, you know, roll out more enterprise use cases, I think employers are excited about the efficiency, but they're also, you know, wary of like what is what are what's sort of being unleashed. >> Yeah. >> Maybe even on the other side, have you guys seen the response from application software around, okay, now there's computer using agents accessing my legacy workflow software, and how do you do you guys think at all about that? Are you more focused on just the the agent that's being used by the, you know, consumer enterprise user?

>> Mhm. Yeah, I think this like I think this problem becomes really interesting on the open web, actually. I think when it's like within software that is like approved and used by the company, it's actually more straightforward because it's like, oh yeah, 100% we want to have uh people automate and do work in here that they don't have to do tediously themselves. But when it comes to the internet, I think you kind of have a few different camps of folks, right? One is uh we run of our business model is not in line with having agencies our product, right?

You think about advertisers, publishers, whose business model's entirely on advertisement. If the agent's using the website for them, then where's that revenue coming from? And I think most places a agent's ad impression does not actually count towards real uh revenue. Um and so that's a good interesting question that we're always working through and working with our partnerships team on. Um there's another question kind of around um for the companies who do really want to embrace having computer use agents using their sites, a lot of them are adopting kind of standards that let humans or people and agents be able to use it together. So like Web MCP is a standard that we're really excited about. We we add a support for it in the app browser.

And so you can kind of imagine these scenarios where the agent is elevating the experience of using the site, but it's not entirely replacing it. So like I think browsing the internet is still very fun, in my opinion, right? I still like doing shopping, I still like looking at all the hotels and flights and places. And so, you know, you You can you can imagine places where um the the agent does all the tedious work, but it kind of carves out an user experience where you're doing the browsing. You're like, "Okay, now I want to finish the the transaction and then put in my credit card and do all this stuff." Then you can hand it back to the agent. So, WebMCP is kind of an interesting way where your interface with the website is still the site itself versus an API, but then the person is kind of using it together. And so, those those are some thoughts, yeah.

>> Shifting gears, want to double click a little bit on what it means to do the job or for get the end-to-end job done. Um would love to hear a little bit from you guys around, you know, what works well, how do you guys define the right the right narrow repeatable task, uh you know, and what sorts of tasks you think um are ready to hand over to the agent versus the ones like, "Okay, it still requires some checking and correcting as the um the user side." >> Mhm. I mean, generally, I found the model to be pretty capable to do anything that you can do on a computer now. Like and and you know, one exercise that I try to force myself to do um is that whenever I have a task, the first thing I think about is just can I just have the agent do it? Now, there are obviously some sensitivities around, you know, if I'm sending a message to someone, I want it feel like it's coming from me. Um but in general, I think the way I think about it is the computer using agents now since Astra um is now capable of doing anything that a person can do. And uh there are questions around, I think before of

like, "Oh, there are some like squishiness around like specificity and requirements." I've I've generally found the model's getting pretty good. Uh whenever you're asking to do something and there's some vagueness, it will actually, you know, with Astra, I think we launched this like uh flexible input, right? And so, it will actually ask you about, "Hey, like I want some details." And if you don't answer it in time, it'll just keep going and then pull the context as necessary. I think there are some things around, you know, if you have repeatable workflows, it's nice to be able to kind of define, you know, the the requirements or at at similar to like writing a product spec. Um, but in general, I I generally feel like unless there is something incredibly sensitive that you want to be in the loop for, I I just reach for the computer use model for everything I I do now. I'm curious if it's the same for you.

>> I Yeah, it it it is I'm I'm trying to uh you know, build that impulse to to reach for it because you know, it's so good and sometimes it's so easy to get locked into the old way of doing things. Um, I um, I think the really cool thing about computer use as a technology is the universality of it. You know, it works with every application. It's unlike any other thing that we've built where, you know, the developer has to do some work to adopt it. Someone has to build a connector MCP server. You know, computer use works with everything and it works by using the application the same way a human uses it. Um, which I think is something worth understanding if you haven't spent much time with this technology. It's like it computer use uses computers the same way a human does. It makes the agent have all the capabilities you do. Um, which I think is so powerful.

>> Yeah, and I guess like adding on to what you're saying, like part of it is like I think over time I've stopped thinking about it as like am I using them on like a computer use task or a non-computer use task? Frequently they're like connected, right? Like the way we think about computer use is that like computer use is not like a thing you pick up and just say I want to do something with it. It's really like the universal kind of API that the model has to do whatever it takes. And so if you have an ambitious task and you want the agent to do it, likely the more complex it becomes, the more likely it will probably touch computer use at some point in that journey.

Um, and so like maybe the only time that I I don't reach for computer use for some cases is if a task is so simple that then it's like easier for me to just go and like click two buttons and do it. But you know, as computer use gets faster, even faster, then the question becomes maybe I don't even need to do that. I just, you know, drive voice and it just uses the computer with me. So those those are some interesting thoughts there, too. Yeah. >> Do you think there's anything you have to share with like an enterprise user to get them comfortable with that?

Obviously, you guys are totally filled on computer use, but you know, I think there's obviously like hesitant sometimes on enterprise user of like, okay, well, yes, the human would click through it and probably do it worse, but there is that feeling of like there's accountability and and you know, uh is it that person's trained and you know, all the all the classic kind of rebuttals. I'm curious. >> Yeah. Yeah, we've had to convince people around For sure. Lots lots of conversation enterprises kind of over the last little bit and I think the the main thing is we think about it as two layers, right? The first layer is like, how do we create deterministic controls?

I think that's really what gives um you know, enterprises a lot of feeling of safety is when they can even though our model is the most aligned ever and it's I think better, you know, better than most people at you know, protecting against prompt injections and unintended things because it's like just, you know, the intelligence so smart, but um I think there is just a feeling of safety when the enterprise admins can basically say for certain task or for certain data sources just too sensitive to be used. Um and so we see kind of enterprise this what we we launched with two things, right? One is the ability to be able to deterministically create a sandbox around your agent. So being able to say it can use certain apps, it can't use other apps, it can use certain websites and not use other websites. And then two is, you know, this kind of part of a human in the loop. Like what type of actions does it require a human to um be intervened? And so you might say, okay, for anytime there's like a download and upload, a person has to kind of approve it. Anytime it goes to a new website, a person has to approve it. Now again, these are all like dials and enterprise

can kind of go through. Um and so it's really like that's one and then two is really just showing the enterprise admins just how you know, all the evals that we've run internally around how aligned these models are even compared to like, you know, both internal and kind of third party know, Evals. I think there's also some baseline stuff that we do across the board to just try to help people build trust with these kinds of products. You know, we know it's so new. It's such a magical and sometimes alien experience for people. And so, you know, we put a lot of work into the design of the product so you can always see the cursor if you want to watch it perform a task. And it sort of animates in a fun way so it's not boring to watch. And then, like James was was saying, this is very con- customizable for enterprises, but also for a regular user. You know, we ask you for permission before using a new application or website, depending on which sort of modes you're in. And I think that helps people sort of not feel like it's going to go do some runaway stuff on their computer.

You kind of want to have this trust-building mode moment where you figure out what it feels like and what you want it to do for you. And then, and then you sort of feel more comfortable over time. >> The Cognex cursor is so cute. I love it. >> It's like such a I know it's a small detail, but like watching it use the computer gives you like confidence and it makes the experience nice. So, I think there's a kind of like a user experience point that's that's worth that's definitely worth investing on it. I guess I'm I'm curious like in a broader sense like, okay, now we have like capable models Mhm. They're getting like cheap enough to run at scale.

In your view, what should kind of companies think about automating? Like what would you What Where do you see the most promising kind of use cases within enterprises? Are there some workflows? Maybe something around the ERP? Maybe something around kind of how, like again, humans and agents can work together to get tasks done. But like, if you had to like pick, where where are you most excited to see your models perform in the world? Like >> Yeah, I guess like from my perspective, I'd love to hear Ari's. It's like two two I think of it as in two big categories. The first category is like automating, you know, tedious workflows.

That's like the more boring one and I'll give you some boring examples like >> Boring boring is opportunity. >> Yeah, [laughter] exactly. Like ERPs is a really big one. And so all the all the acronyms you can think of ERPs, CRMs >> All the systems of record. >> Yeah, every single system of record, right? And so we know we've seen folks, you know, procurement software. Like, you know, you are going to vendors and every year you're ordering a bunch of lab equipment. Instead of having this giant spreadsheet with like a thousand rows and then the ops team sits like, you know, worked with one company it's like normally what they have done is you would they have this giant spreadsheet that gets filled out by the teams themselves. Then an ops team takes that giant spreadsheet, puts it into an internal, you know, procurement software and then that gets like, you know, finalized as a bunch of lines and then sales team then goes procurement team then takes that giant software output and then goes out and reaches out to vendors.

Computers can automate all of that, right? So like procurement is a big one we see. Finance is another really interesting category we've seen. And so you know there's a lot of different ledgers within a company, right? And so you have like ledgers for expenses, ledgers for all these things. And so at the end of the month there's a lot of reconciliation. And a lot of these us financial software don't have APIs to be able to quickly pull things. And so the finance team being able to have computer use look through all these ledgers, reconcile, pull out anything that's like interesting to flag either to a person or then do some investigation. Also a very boring but useful workflow. From our sales team, a lot of sales teams uses a lot of different software for like transcripts, for getting data, all these kind of things. And so having computer use go and do prospecting and then going and putting that into a CRM or some kind of like, you know, outreach software and then all that. So all the system records is the boring part.

And then like market research is another one that we see a lot. Um, and so, you know, in my free time I've been building this like uh video annotation software for for badminton to play a lot of badminton. And like I like to film myself playing the game and then uh having the agent kind of go and and look at stuff. And I was like, "Okay, how do I make this thing really useful?" There's some software out there for like pickleball, for squash, for tennis. And so, I just had computer use go log in, create an account on these on these softwares, try it, take screenshots, give me feedback, and then try to blend it back into the app. And so, these are all kind of like tedious workflows that could be done. And then I think the some of the more like interesting creative work of like enabling more ambitious work is like, you know, we talked about um the model now being really good with like visual visual design. And so, um circuit design, you know, being able to like architectural work, hardware engineering, game design, these are also areas that I I think we're really excited about where Astra can take. Um, less, you know, less common for enterprises to think about those right

now because they're thinking about the boring opportunities, but I think as folks understand the capabilities of model more, that's where um I think we'll we'll kind of see more usage next. But curious if you have any thoughts, Ari. >> Yeah, I think kind of similarly on the consumer side, there's a lot of things where it's sort of automating drudgery or just doing things you don't want to do. It's filling out DMV forms for you. It's helping you uh unsubscribe from a bunch of emails in your inbox. Um, and then I think there's also things that are more like um I'm I'm I'm hearing from a lot of people stories of people doing things that they just wouldn't have time they wouldn't have had time or wouldn't have thought to do otherwise. Like, I don't for me personally, I'm um uh maintaining spreadsheets about my finances, spreadsheets about meal planning, and uh I have a friend who's a personal trainer who's making customized things for his clients. And um you know, some of these things um would just be so so uh time prohibitive to do in your in in in your personal life. Um, but this makes it so easy. Uh I use Apple Numbers, which is like a obscure spreadsheet program, but the model still knocks it out of the park, you know?

>> heard someone use Apple Numbers. >> It's a surprisingly great spreadsheet software. Um and then I think there's a third category of thing which I'm kind of the most excited about, which is just starting and I think uh uh you know, on the horizon, you know, um which is really just having a different way to use your entire computer or interface with software. I think you saw in our Astra ad really cool examples of people just talk to their computer with their voice. It's such a natural way to use your computer and I think over time we're going to see the form factor of software itself change due to this technology and I'm I'm really excited about that.

>> On that point, okay, form factor of software changing. Um imagine if you're an enterprise user, you've used to you're used to using whatever clunky ERP from was built, you know, 30 years ago. Um there's a world in which you now, you know, unleash uh a set of agents that just operate that software and you don't really care. Um there's another world in which um folks are a new like AI native player is using computer using agent and I'm curious if Do you think the world is going to go more in the direction of like keep the legacy folks um and just have this layer on top or do you think you know, computer using agents can absolutely be part of uh new software players? And this is something I think a lot about, so I'm curious to get get your guys's thoughts.

>> It is such an interesting moment where simultaneously the cost and complexity of creating new software is so much lower, so you can imagine software changing so much, but also there exists so much legacy software in the world that computer use is so well suited to interact with. So, I expect both of those things to continue for a long time. You know, if you look at the software that's in production today, a lot of it was written 10, 20, 30, 40 years ago and I I I don't think that that stuff's going to go away overnight. I do I that in some way the existence of computer use uh puts may put more pressure on companies to make software work better with agents because all the software is going to work with agents by default and I think WebM CP is a great example of this and I think we'll see more stuff like that. Um so yeah, I sort of expect both to be to be super important but I do think it'll like kind of shift over time.

What do you think James? >> Yeah, I I I agree with you. I think it's like um the there the the way there's like a lot of arts and craft and you understand this cuz you already built a beautiful product with his team. The cursor was all him. Um and so the um I think there's like a lot of art and craft into building like great products that people can hold in their hand and um I don't know like for me if we go if the internet becomes just this like faceless you know, thing that is just agents talking to each other and then people are not using anything anymore, I would be very sad. Um and and I don't think it will be like that because like there's like something about craftsmanship and artisan software that's like just feels nice to use because the person who designed it really really understands um like the experience that the customer wants to have. So like my wife loves is recently got into playing this like uh you know, board game. I I forgot the name of it. It's where you connect like five pebbles to like win. I think it's

like O Mac is the name of it. And so anyway, uh so she's trying to get good at it because she's playing online with friends and it's getting her ass kicked. And so I was like, all right, I'm going to build you like a tactical trainer for um you know, going through different gambits so you can like get better at it. Like that's for example something that like I you know, created first version. I was like, oh, I know like she likes to do it in a certain way. She likes these type of uh ways to learn more than others and so I was able to like you know, craft the software to be something like feels really good for her. And like these kind of personalized software I know it's like a thing maybe today still feels like only really tech people will do, but I think if you like think about this a few years out, it's like if anybody can build anything and they can build these really craft software in a very short amount of time for like niche use cases that are like so well tailored for people. I think that's a really beautiful world. Um and again, you know, at the end of her training, the lessons on mistakes she

made, that could be an agent that summarizes it and gives you her insights format. But when she's actually playing with the with the AI and and and and and trying to get better, that's something that the person has to do. So I do think it's it's I I I'm more I'm bullish that I I don't think UI's going away. And and uh and I think kind of with low tangent to your original question, but I I do think that um the pressure of cheap software will make it so that software has to be better. Um but I don't think it would just be like yeah, everything will have an MCP and an API and you know, computers will never be useful.

>> Totally. >> this point so much. I think that like humans are so hardwired for direct manipulation, you really want to be able to you know, move and and and do things with your own hands. Um and I I think that it's really exciting to think about how software in the past was so one-size-fits-all and had to be, you know, often uh a lot of the software we interact with every day is designed for millions, tens of millions, hundreds of millions, billions of people to use. And I think we will enter an era where any piece of software could be designed for a hundred people or one person. And um just how much more uh personal and and and unique that kind of software would be.

>> Yeah, I my mental model is I think there's probably two buckets. One is on the consumer side, you've got um either something [clears throat] you just get you you get joy out of you enjoy using or you learn from. There is this it's not the utility of it, it's the joy of interacting or learning or whatever it is and you want to spend time. On the on the B2B side, you you may want your agent to complete a bunch of the tedious work, but you also there's this concept of like the audit log. Like and you that could be personalized in how you want to interact, but the product really is me as the human user setting the goal and then being able to like understand the logic, but not necessarily need to run through like the entire chain of command of how it happened. And so like any B2B user still needs to figure out okay, like why did that payment get made and go back in and get a summary of what happened and be able to say okay pay this person at this time. Um yeah, at 30 days this this much discount should be applied, etc. They want the agent to take care of like the

execution of it, but they want to be able to at some level like some level of control still needs to be put put in there and I think that's like something that does require a UI of some sort. And it has to be I think if our if we're moving more towards believing that AI will do more of this and that part has to be enjoyable to use, too, right? Why not? Like why should we settle for shitty software? You know, it's >> I do think it'll be really interesting to see if we and I I'm optimistic that we will see new kinds of software where it has traditional user interfaces, but is also using agents under the hood and then maybe sometimes those agents are themselves using computer use to automate legacy software by giving you some much nicer, more purpose-built experience. I'm like really excited for for that class of thing.

>> Yeah. >> Yeah, it's kind of like the one where we like to use in terms of the shared intelligence layer, right? Like you have your agent using legacy software somewhere else and doing things other places, you like if you're using an application, the application wants to get access that context so it's able to deliver a better experience. Like I think if we had every website has its own agent that doesn't talk to your agent, it's it's a very fragmented world and so figuring out like how that plays together where the agent can plug in and and use your other software, I I think it's going to be like a really interesting thing to play out for the next few coming >> months.

>> I love that you guys, yeah, talk about this cuz I I think it's so underappreciated the importance of like human AI interaction, especially if you want like a product to be successful. So, like being thoughtful of like how can I kind of like humans are used to collaborate, so you want any agent to be able to like kind of fit into the team. I'm just curious like if you apply that to a new startup, let's say you're like, I don't know, starting a company tomorrow and you want to use computer use to like solve a workflow within the enterprise, how would you think about it and would you be kind of what would be the the things that jump to mind in terms of like, oh, I want to know when it fails well, I want it to be able to communicate with people, or I want it to be like I don't know, a black box just like maybe if you had to talk to founders right now, what advice would you give them as they try to use your models and capabilities to just solve problems for enterprises?

>> In my mind, it depends on it depends so much on the modality of, you know, are you trying to are you a small team and you want sort of individual on the team to be able to use some piece of software that that your that that your team is using but in a more efficient way or are you maybe building something that itself wants to interact with existing software, in which case maybe you have some, you know, machine in the cloud somewhere that's doing computer use in the background. I think that will really affect the modality of how you want the human agent collaboration to be, right?

And yeah, you know, in the former case, I think a lot of our products are serving this use case, you know, where you sit down at ChatGPT and you know, there's a lot of great you know, collaboration you can have there. And in the other case, you know, maybe you're building something with with the API, you're you're doing doing your own computer use implementation and then yeah, then it really is interesting to figure out when when do you want the human to be in the loop is the agent telling you when something's wrong is it telling you when something went right you know is something summarizing that so that you don't get bombarded with notifications saying hey this happened this happened this happened which actually you're going to ignore um um I don't know what what do you think James >> Yeah I mean I I you know both Ari and I were startup founders before we went open AI and so you know I'm I'm still very bullish on startups you know I think with the question right kind of comes down to like what is there still a moat right is there something that startups can still tackle while labs are are doing things I mean in general I just think there's like like two two things like one is like I think there's so many things to build right

and like even though we're quite big now thousands of people we are still feeling very stretched thin internally and like you know it's like the teams are not very big for all the problems and so if you have a problem that you really deeply understand you have industry expertise you understand how regulatory works the go-to-market motion how to service the those customers it is still I think a very green field out there and um the the tools that that you have through our APIs like Ari mentioned with computer use I think I think we'll still let you accelerate and speed really I think at the end of day speed and pay is kind of moat that that exists today I think the other thing I think about a lot is like there are in my opinion still a lot of like vertical industries where the diffusion of AI is like much slower and so you know I was talking to a company the other the day who uses computer use to help restaurant groups to you know take pictures of like their back storage you know and then using that the the vision

looks at kind of like what is going to be missing based on a projection of tomorrow's traffic and then it takes that to then use computer use to go to like a Shopee or some kind of like procurement software to then order your groceries for next day and then based on that it will go into your ERP to adjust like the timetable for your like employees who are um who you might need more people to help because in the last year at this time of the day, there was a lot more traffic coming in from like like holiday or something like that. And so like, you know, those restaurant groups are probably not like tech companies that are super, you know, familiar with all the tools that are available. And so I think like, you know, living in San Francisco, we see a lot of thinking, "Wow, like the labs can do everything, the codex and like why doesn't everyone just install codex and just have it do stuff?" But like for the average American and the average small mid medium business owner or the average enterprise, even understanding what the tools are capable is one, connecting tools in a way that makes sense is two, deploying

it is three, servicing is four. And so there's still a lot in my opinion where if there there are complex workflows that require a lot of disparate software, computer use is a great way to join them, complete the entire end-to-end workflow, and then building it in a format and UI that is feels good to hold for those customers feels still like a really really big opportunity that, you know, we I think is still open. >> Tell us about what's next. What what are you excited about that's on the frontier that you can share around around computer using agents?

>> Well, I've I've sort of gestured towards it already, but we're really excited about speed. We're really excited about making computer use superhuman and we're really excited about things that blend together different kinds of interfaces that bring together sort of multimodal inputs so you can have more kind of fluid experiences with your computer. I think right now a lot of our computer use product is very background focused. It's very much like let me give the agent a task so it can do that while I do something else, which is actually a tremendous strength, right, of the product, but I think that in the future we'll also see more of this like real-time collaboration between you and the agent.

So really excited about this area. And um Um, other otherwise just sort of better, cheaper, more more available, easier to use. >> More Yeah, exactly. Being able to actually access everything. >> Yeah. >> I think I think there is you know, we with dots, right? Where we're giving folks these like computer in the cloud. ChatGPT work also has a browser in the cloud. I think it'll honestly like for for me like one thing I'm excited about is just making the experience feel a little bit more more coherent. Like, you know, today you can install the Chrome extension and it that uses your like password manager and everything in Chrome. And then we have an in-app browser in the desktop app which has a different set of cookies and different passwords. Your computer in the cloud also has different things. And for the average consumer, it's kind of confusing. It's like, why do I have these four different browsers that all don't talk to each other? If I open up a tab in one, does it like show up in the other? And so like I think a lot of what I'm excited about is how do we make computer use go from this product that is used by like enterprises and like

really hardcore tip of the spear kind of users to something that's actually understandable for for people. It's kind of like an oxymoron because I think computer use is actually the most easy to understand AI product because >> Right. >> it's an AI using your computer. But then because of the like uh complexities of like the different like states you have to manage and the different places you're using it, the implementation details sometimes seep through and then it becomes kind of hard to grok. And so um being able to kind of make that consumer grade and easier to understand, more more coherent um is something I'm really excited about for next quarter.

>> Yeah, I think it it comes back to the point that there is still a human that's setting up up the agent, right? And they have to the the the better the UI and controls and the easier and more enjoyable it is to use, the more they're going to get out of it and that's >> Absolutely. >> On that point, one other thing I want to gesture towards is um I think another challenge of the product today is you you have to put some thought into how you're going to use it and you have to remember to use it and you have to um convince yourself to do things that are different from how you're used to doing them. Um and I do think um dots like James was talking about earlier um have this really powerful proactivity where they can sort of come to you and identify situations and and and and um and and help you solve problems. I think we'll also see a lot more proactivity.

I'm really excited about a world where the agent can sort of adapt to you instead of you adapting to it. Um and I think that we're going to see a lot of that and I think it will um make it a lot easier for people to actually like relate to and uh and and understand and incorporate this technology while also um still being the one sort of setting and steering the the the the goals if that makes sense. >> One of the things we haven't talked about yet is the momentum behind computer use and how much has changed. And Fabrizio and I chatted a little bit about this in in our piece. Would love for you to just kind of talk about where it's you know, how it's the trajectory's so far and where things are headed.

>> Yeah. I mean James and I have been working on computer use for some time now. >> Yeah. >> Um at Open AI and it's been an incredibly exciting past like 6 to 9 months, right? Because we started working on these computer use products and um very few people were using them. Um the models um all of a sudden it felt to me became great at these types of tasks um earlier this year because of amazing work the research team did. And then we've been working super hard with a bunch of teams internally on how to create actually good product experiences around this. And then just in the past couple of months all of a sudden, you know, the product's growing like crazy and and people are really seeing the value in it. Um but I think it's still so early, you know. It's still the case that only um some you know, portion of our user base is using computer use and um I think we're going to see it uh grow quite a bit.

>> Yeah, totally. I mean, you know, I joined Open AI in January and that was I joined to work on the Alice product. And love love Alice. I wrote Alice. But the uh the I think, you know one of the things that we realized was um in the Alice product, the agentic experience, I think for a few reasons wasn't actually like a big selling point. Not many people used it and partly because I think people mostly treated it like a normal browser. Um and over kind of the last few months because the model capabilities have gotten better, we've now incorporated browser and browser use into kind of all of our existing products where there was already agentic usage. We've we've just seen you know, it's like YC level of growth week over week. And then I'm always surprised by how fast it can go.

And I think this next like big inflection point is like cloud, right? You like how do we like desktop is still very much a power user thing. I know we probably use our desktop all the time, but like most most consumers are on their phones. >> Right. >> And so um how do you give every single ChatGPT user, how do you give everybody who's just using their phone for their day-to-day life um a powerful computer that they don't even need to understand how to use that it just kind of makes things work for them and I'm really excited about how that kind of brings it to, you know, billion users.

>> It feels like computer using agents have had a ton of momentum behind them and now it's the moment of like, okay, how do we get it into more people's hands? They trust it, they feel like they can really use it and to hear your experience your point earlier about coherent experiences. >> Yeah. >> Like that's that's a real I think once as that gets a lot better, I think consumers will feel more comfortable using it and and it'll feel more natural. >> I think that's right. I think I think we're at the skilling moment of computer use both for >> consumer and enterprise >> Yeah. >> at the same time.

Um and I'm yeah, really excited to see where it goes and what people build with it. >> Nice. Us too. >> All right. >> Well, thank you so much for joining us. This was awesome. Um we love talking about this topic and it was really fun to hear it from you guys directly. >> Thank you. >> Yeah, a lot of fun as well. >> Yeah, thanks. >> [laughter] >> Thank you. >> Thank you guys. >> Mhm.