Transcription
Maybe the right UI is a generative UI. Imagine the entire layout of the workspace is actually something made just in time, based on what you're trying to do. Cove is now going to be building out the workspace, and so as it builds out the workspace, it's trying to first understand what you're trying to do, how to break down that problem, and how to help you make progress. That includes, for example, for YouTube, being able to automatically get to the transcript. As product builders, we're trying to kind of predict the future, and that's already hard, but the future feels like it's changing faster and faster now. Stephen built Uber Eats from scratch, uh, of course with his team, to a $25 billion business. The initial group that worked on Uber Eats was a really interesting combination of some folks that had been at Uber for a very long period of time, some of the original ops people, some original engineers, and they were mixed with a new set of folks on, for example, the product and design side. That mixture of new perspective, but also folks that really knew the Uber culture and really knew how to get things done at Uber, was a really interesting kind of like cocktail for an initial team.
Well, welcome everyone. My guest today is Stephen, the co-founder of Cove. Cove is a Sequoia-backed startup that lets you work with AI through a groundbreaking visual interface, instead of using chat threads and and before Cove, Stephen built Uber Eats from scratch, uh, of course with his team, to a $25 billion business. So I'm super excited to have him demo Cove and talk about how he's thinking about AI beyond the chatbot. Welcome, Stephen.
Hi, thanks for having me.
Yeah, so uh, why don't we get right into it? I would love for you to show us how Cove works. Uh, I don't know, maybe we can do like a travel use case, like I'm I'm heading to Japan soon. We can maybe show how Cove works in that context.
Sounds good. We'll try to plan your trip for you.
So let's do it. All right, so um, so this is Cove, and what we'll do is we'll go ahead and tell Cove what we want to work on, and so you said you're going to head to Japan soon, right? And like what cities are you going to be going to?
Uh, yeah, I'm going to uh Tokyo, and then uh Kanazawa.
Uh-huh.
And uh Osaka.
Okay, with my family.
All right, cool. Help me plan a trip to Japan. Say I'm thinking about going [Music] to Tokyo, Kanazawa, and Osaka.
All right, cool. So Cove is now going to be building out the workspace to try to help you figure out how to deal with this task, and so as it builds out the workspace, it's trying to first understand what you're trying to do, how to break down that problem, and how to help you make progress. So let's see what it does. All right, so what you see here is a couple different cards. We're going to close this for a second because we're going to talk about that feature in a bit, um, and so we have a couple different cards here, and so the first is a card that is showing some of the cities that we mentioned, and so it is showing kind of a comparison across some of those cities because it's thinking about how you might be trying to make a decision over how much time you want to spend there or the case is, and then it's also starting to build out an itinerary for you as well. And you'll see that there's a card here that actually is trying to almost elicit different preferences from you because Cove may want to know different pieces of information before it fully um is able to kind of give a good job of giving you that sort of itinerary, sort of plan. So we'll fill out some of these things in a bit, but maybe I'll first show some other things in terms of the product. So Cove is a visual workspace, and so um you can think of it as a canvas, and we're going to be able to explore a lot of different things as we work on this. In addition to that, everything in this space is fully editable by you. I feel like for folks that are used to working with a chatbot, you're used to kind of looking at a chatbot's answer and then needing to kind of copy-paste that or watch it kind of rerender as you kind of make edits. Cove is different. I can come in here and I can just change any of this content automatically, and I'll show you how that works as we kind of go.
Um, so why don't we add some more content to this space? So, for example, I'm going to drag in a new card, and then let's say um say let's a table of kid-friendly museums in Tokyo. We're both parents, right? So I think this is probably something that might be might be helpful, and so I could just start to create that content myself. Cove is also always available to help fill it in. So I clicked fill, and so now this table is being created for me, and what you'll find is when you're working in Cove, what we're trying to do is almost kind of create the experience of how you tend to collaborate with other people, right? So if we're working on something together, we'd be iterating together, building on ideas together, and so similarly you can do that with Cove as well. And so let's say, for example, we wanted to add some photos to this. I'll just say add some photos to the table, and then it'll go ahead and do that.
Yeah, this this is awesome. Um, yeah, it's kind of like I guess like it's kind of like a Figma, but you know I I guess AI is generating a lot of things with your input, and you're working together too, right?
So as you noticed when I said add some photos, it didn't rerender the entire table, right? A new column was added right there, and now it's filling in tables, and we just feel like that's a much more natural way of working with AI where you're co-creating, not needing to kind of see all this content rerender. So this might be on your list, Ghibli Museum. My kids are so so for example, that might be something that you're interested in going to. Um, and so and then a couple other things to kind of call out. So for Cove, as you add more content to the space, the AI is always trying to understand the context of what you're working on, the information in the space, and how those things come together. And so one of the things that it does is it tries to anticipate other things you might want to do, and so anytime you have a piece of content, you can see that there's actually suggestions here over other things that you might want to do with the content. So you might want to get links to the museum websites, you might want to add more museums to the table, and so it's really easy to just continue to iterate on things like really quickly.
Um, let's see what else should we talk about here? Is it is it kind of like uh because uh with a chatbot you only have a single thread, right? I guess you can go back and edit your previous prompts, but but like this feels like it's like a multi-threaded thing, or like it's kind of
That's right. That's right. Yeah, so you can do you can invoke the AI to do various things all in parallel, and those things will all just be happening in the space. And then the other thing to keep in mind is like right now I'm the only one in this space, but this can be a multiplayer space. You could come in here as well, you could be doing things at the same time, and when you're here with me working together, we can both invoke AI at the same time, so you see some like really interesting things that happen as people start to collaborate together in these spaces.
Oh wow.
All right, so let's um maybe we'll start to kind of build out an itinerary, right? So as I mentioned, there's a card here that actually um is able for you to kind of put in some additional kind of personal information. And one thing that we found is as you work with AI, sometimes people don't know exactly what's the kind of additional context that you might want to like provide AI for it to be helpful, and so we actually have these cards where it actually kind of helps give you hints of like, oh actually if you provide us information, then AI can actually be more assistive in terms of what you're doing. In addition to that, when you fill in this content, sometimes this content can actually change other cards in the space automatically. So for example, let's say that we're going to go for seven days, um spend most of the time in Tokyo, but stops in Kanazawa and Osaka, and say you have a chill travel style. Main interests um we're going to do nature uh nature.
Yeah, food definitely.
Oh yeah, food.
Yep, food. And then we'll say kid-friendly activities, maybe.
Yeah, somewhere they can babysit my kids. No, I'm just kidding.
All right, cool. And so anytime you start to do more things in Cove, in this space, it's going to take this information into account as it helps you, but in addition to that, there's actually this button that says do updates, and so it actually has connected this preference information to this itinerary card, and so if you click do updates, and it's going to actually automatically update this itinerary based on that new information. And so as you continue to explore, as you continue to learn, you add more context to the space, and things are able to continue to update based on what you decide, so you're able to kind of get to the exact um in this case itinerary for the trip that you want to do.
Yeah, it looks great. It looks like it first first visit is the Ghibli Museum.
That's right. That's right. Yeah, so so it's it's using the context of of of that card that we created, and you know context is something that we think is really interesting in terms of just how you can use that in this product because there is let's call it implicit context, right? So an example here is it we learned about the Ghibli Museum and went ahead and started to use that when it created the itinerary, but you can also direct it as well. As you start to add more cards to the space, you can actually mention cards to do specific things for cards connecting to each other, and so there's a lot of like really interesting kind of workflows and things that people can start to building Cove as well. So you can uh basically so instead of like one input box, now you have each of these cards, you can add more roles or add more inputs, right? Until you add cards, and also there's many different types of cards that you can add into the space as well. So you can add URLs, you can add PDFs, um you can drop in YouTube videos. All these things are additional ways you can kind of build context and be able to use that that content in some sort of way as you work on things in Cove. And and uh with chatbots, I I normally have to like copy and paste it to like some Google doc, but you're saying that this one you can just share like I I want to show my wife, that's just share it.
Yeah, yeah, exactly. So just like um you know a lot of the other kind of multiplayer collaboration products you're used to doing, Google Docs, Figma, whatever the case is, uh you can share this out to particular people, or you can make it read-only to uh and share out a link, um so you can have kind of like all that sort of collaboration uh as you work with AI. Um you can also download these cards, you can actually move them into other places, and so we we see that to be pretty popular as well, as people are either creating presentations or other things are able to like kind of create use the content that they use in Cove and actually move it to other places too.
All right, let me give you the stress test. So like uh can you ask it to can you ask it to like list the top whatever 10 10 uh snacks in Japan, like ah okay 10 premium premium snacks. I don't want Pocky and stuff. This episode is brought to you by Merge. Product leaders cringe when they hear the word integration. They're not fun for you to build, launch, or maintain, and they probably aren't what led you to product work in the first place. Luckily, the folks at Merge are obsessed with integrations. They built a single API that helps SaaS companies launch over 200 integrations in weeks, not quarters. Think of Merge like Plaid, but for B2B SaaS companies. Like RAMP, Drata, and Electric use Merge to access their customers' accounting data for bill reconciliation, file storage data for searchable databases, and HRIS data for auto-provisioning access for their customers' employees. If you need AI-ready data for your SaaS product, then Merge is also the fastest way to get it. So if you want to solve your company's integration dilemma once and for all, book a meeting and receive a $50 Amazon gift card when you attend. That's merch.dev/peryang. Now back to the episode.
Yeah, so um so I'll do that. One thing I I'd point out here is that there's also um you know Cove is always trying to anticipate how you might like something you might want to do next, and so as I come in here, if I don't type in anything to the chat, I can also see recommendations of other activities or other um pieces of information I might want to dive into. So I can click on any of these and also create a card, but we're going to focus on your snacks. So let's say uh create a card. Yeah, I want pictures too. So you add pictures. Ah okay, okay. All right, create a card with pictures of premium snacks Dubai in Japan, maybe by region.
Sure.
All right. Yeah, this is my way of uh making my wife happy, you know, I wish she the fine dining restaurants and stuff, that that'll break my bank. Bring an extra suitcase so you can uh fill it up with this stuff.
Oh nice. Yeah, I haven't heard of some of these. Oh, Tokyo Banana is really good, so so I definitely suggest that. And where where's it getting the images from? Like just from
Yeah, so there's various providers that we use in Cove for all the different things that you're seeing here, so um we there's you can do web search, we have images um and various other things as well. The images here um are actually through Bing. It's actually like the Bing API from standpoint just happens to be really easy to use, uh so we're using them at the moment.
Finally, I finally use for Bing.
Yeah, okay. Yeah, this looks good, man. I think uh I definitely tried some of these before. The white chocolate ones are great.
Yeah, the the the Japanese have a you know very high craft for their snacks.
Yeah, absolutely. Best part of the airport is you go and you can load up on samples and stuff. So yeah, so so I I noticed you have uh on the left here an AI app feature, so so what is that?
Yes, so this is something that we just recently launched, um and so I can show you a demo, something we're really excited about. So maybe I'll show you a different space, and so this is a space where um imagine you are putting in your class materials because you're studying, and so this is actually a very common use case that we see in Cove. People will put in lecture notes, um you know like lecture videos, they can put all into the space, and then all that becomes part of the context of the AI, and they can do things like ask questions about the content or summarize it or whatever the case is. Um so now with AI um with this new feature that we launched, you can actually build AI apps right here within Cove, and so I'll show you an example. I'm just gonna drag out this card, and then I'll say something like uh create a multiple-choice quiz about, and then I'm going to go ahead and show you how app references work too. I can just add mention this video, for example, and so this video has has already been transcribed um because I dropped into the space, and then now I'm going to just click build an app, and then it is going to start coding.
Well, you just dropped a YouTube video into the space, and it's it transcribed it.
That's correct. Correct. Yeah, so so um so the you can drop in videos, PDFs, um you know word documents, and so all of that just automatically gets um added to the the context in the space. And that includes, for example, for YouTube, being able to automatically get to the transcript. Um so all that is um stuff that you just get as you add content into Cove, and then now you can actually create these AI applications, and because these AI applications are in the workspace, they can actually use the information in the workspace as well. And so the example here is that you can actually take that lecture transcript, and then now we're going to create a multiple-choice quiz on it, and so this is Cove coding away, and so this is going to take um you know like maybe a minute, and then maybe as we do that, I'll show you something else fun. This is another app that's in this space, so imagine this person's studying physics, um you know there's a they're learning about the three-body problem, and so this is just a little visualization of the three three-body problem that we created, and so you can change the different uh parameters for each uh planet, and then you can click play, and then I'll just go ahead and do the simulation, and we have a we have some pretty crazy orbits going on right now based on
Yeah, so this is a this is a app that uh so what was the prompt? Just like build build a simulation, create a visualization of the three-body problem.
That's right. And it created that.
Wow. Yeah. And then so
Yep, so here's our multiple-choice app, so that's been created, so we can uh we can test your cosmology. Let's see what's the primary focus of modern cosmology according to this lecture. I'm going to guess this one.
Got it wrong.
Nice. And and and uh so if I'm the instructor and I make the stuff and you're here as a student, you can see me make the changes live, right?
Ah, yeah. So so you know because you are creating these apps within a co-workspace, then there's a bunch of other properties as part of the workspace that you just kind of get for free, um and so an example of that is uh the multiplayer part, right? Just like how we talked about in the Japan workspace, you can invite other people, and you can be collaborating together. You can do the same thing with these apps as well, so you can come into this um this space, and then you could be answering questions. I could be watching you answer questions, and so you can actually have people create apps that you're able to use with other people, and so that's one thing that we're excited to see, just like the different types of things people create.
Yeah, it could be really fun, kind of like uh like in a Figma jam, if you're doing like a brainstorming, it's just fun seeing other other cursors around, right? Something similar.
Yeah, yeah, yeah. No, there definitely is. When you think about using Cove in a work context where it's teams collaborating, it is really fascinating to think about a group of people working together, and they can just just-in-time create any application to do something together, right? We've we've seen teams create apps to do things like vote on what where to go for their next off-site, right? There's kind of like you know like things like that they can do, or there actually can be like more workflow or other things that they build these apps around. Um so I think it's it's really fun to see kind of this combination of being able to use context within the space and also be able to bring other people together, and then you're able to have these these applications get created so quickly, and you kind of mix all those things together, and it's really interesting to see the type of things people are doing.
Can you show us uh I notice you have a bunch of other tabs open. Can you show us some of these other use cases?
Yeah, yeah, for sure. I'll show you um this is this is a fun one because it's a little bit more of a a personal one. This is actually my uh my daughter's space. So my daughter is 11 years old, and she um always has some sort of project going on. The current project is she's setting up an aquarium, um and so for folks who have not done this before, and I count myself as as as one of those people, there's a lot of things that we didn't know you had to do to set up an aquarium, and one of the things is you can't just put water in and just throw fish in. You actually have to do this thing where you cycle the water so the water can actually be ready to support um fish. She actually wants to do a shrimp aquarium, which is why she's dropping this picture of shrimp here, and so this space is an interesting one because this is basically you can almost see this journey of this project that she's working on. She first kind of was learning about how you cycle an aquarium, and as she kind of was going through that, then there was this card that got created which shows okay, this is actually the the process that a water goes through, ammonia goes up, then nitrite goes up, then nitrite goes up, then nitrate goes up, and so you basically have to measure the water on a regular basis to see this kind of process happening before you know whether the aquarium's ready. And so as she was learning about it, she's like "Oh, maybe I should just create an app to help me." And so she created in an app to help basically track the progress of her aquarium, and so this is the app she um is able to put in each of the measurements of water, um and then she hits add measurement, and then it goes it when it goes ahead and creates the graph to show that sort of progress, and interestingly it actually even tells tells her what part of the pro the the the cycle that um that the water's going through based on the data. And so this is fun because you know it when the app got created, it used the context of this image, right? To kind of figure out okay, this visualization is actually like really helpful. And as you kind of continue down her journey, um she bought some moss and put it in the aquarium, and then there this these snails started to appear, and then because you know she's like already doing this thing in Cove, and she had this particular problem of like oh now I want to figure out what type of snails these are, she just came to this workspace, created an app, and this is her snail identification app. She's able to ask answer different questions about the snail, and it'll tell her what type of snail it is. And so this is really fascinating to see kind of at when you put the power of creating apps into the hands of different people, and you also put it in the context of a workspace where they're already working through a particular problem, then really interesting things happen in terms of like you know the type of things that they create.
Yeah, this is awesome. Like uh obviously I've used like ChatGPT and cloud projects like for stuff, but uh but I love like in those projects you can only do like different chat threads, but here you can make different apps, and it saves the context throughout, right?
Right. That's great. Yeah, uh your daughter seems pretty advanced, and I don't think I ever want to I don't think I ever want to have an aquarium anymore. This seems complicated. Seems pretty complicated.
Um, hopefully she can keep the fish alive after doing all this.
Yes, yes. We'll see. We'll see. I'm sure there'll be a a much longer journey that happens in this space.
Yeah, yeah. Um let's see what what else? How about the one about Palato? What was that about?
Oh yeah, well, I think this is um this was just to show an example where Yeah, as we start to empower people to create these AI applications, sometimes people may not actually realize that an AI app might be something helpful in terms of what they're working on, and so in this case um
A user has typed in, "This helped me figure out whether I should move from Paul to Vancouver." This is actually like a user we were chatting with recently who was trying to think through this big decision. They're using Cove as basically their thought partner to try to try to figure out whether they want to do it. And so you type in this task, and these cards start to get created. These are things that we kind of showed previously in terms of the demo, but now there's actually a new card that has appeared as well, which is like, "Oh well, if you're thinking about doing this, maybe an app would be really helpful."
And so in this case, it's saying like, "Hey, if you're considering doing this move, maybe you actually want to create an app which will help you figure out the costs of living comparison." And so this is something that we find really interesting because at Cove, we've built other features within Cove where it's all around the AI trying to anticipate what you might want to do and provide suggestions. And that could be, you know, adding another row to the table or whatever the case is. This is a kind of a new version of that, right, which is like, "Okay, actually, you maybe what you're trying to work on an interactive experience would really be helpful, so why don't you consider doing it?" Click a button, we'll make we'll make this tool for you.
"Oh, nice. So the AI is suggesting that I make an app."
"That's correct. That's correct. Yeah, got it."
"How does the AI decide whether to make an app or just make a table or like a chart, you know?"
"Yeah. No, it's it's a really really good question. I think as we have thought about how to design the product, this is one of the things we talk a lot about, right? Because you have particular capabilities that your product can have, and there's a question over: do you—are each of those almost like different buttons and actions that a user is supposed to use themselves, or is it actually the judgment of the AI to be able to figure out when you might want to actually use particular things? And what we found is that being able to provide both of those things is really powerful, right? Um, the AI in many times actually can have really good judgment in terms of when it might want to actually use a particular tool at its disposal. Users also should be able to invoke those tools when they want to use them as well. Um, an example is web search, right? So in Cove, if you do something where recent information on the web would be helpful, then Cove will actually just automatically do a web search to start to get that information. You can also just say like, 'Please, you know, look for look for this information on the web' as well. And so being able to offer both um is something that that we feel like is important."
"In this case, there are other considerations, right, when you think about these kind of AI applications. Um, you know, one is just the amount of time it takes to create these applications. And so, you know, that's one of the reasons why we we have this sort of suggestion here that we show to the user because it is a pretty intentional action, right? To actually build an app, you're gonna actually have to wait for it to kind of code code code the the app before you can actually start to use it. And so in this case, you know, being able to almost provide that speed bump where you actually like are able to suggest to the user, but then actually have them say, 'Yes, I want that,' Um, is something I think makes a lot of sense with the current capabilities of of AI. The thing that gets really fun is to think about how that evolves over time, right? As we kind of approach a a place where the models allow us to be able to code more and more things in a faster sort of rate, then what does that mean in terms of the product experiences that we can build? And so as we have started to kind of build this sort of capability within Cove right now, the definition of these these apps are these, you know, individual cards that appear within the work cont within the workspace. But over time, we actually think that more and more of the entire experience can actually be generative from a UI standpoint. And so you know we'll start to see that sort of transition as the fundamental capabilities of AI coding improve. And so that's something we get really excited about."
"Like uh like like uh it would just generate a bunch of cards or like it'll completely change the whole canvas."
"Exactly. Exactly. You know, like when you like at at its core for Cove, what we're trying to do is try to figure out what is the right UI for AI. And if you think about like, you know, what's the right UI for generative AI, maybe the right UI is a generative UI, right? And so like imagine the entire layout of the workspace and all its contents is actually something made just in time based on what you're trying to do. Now that might sound a little bit sci-fi right now, but I think as the capabilities of the as underlying capabilities improve, like those are the software experiences that we can start to to try to build towards. Um, and so I think we're taking the first step here with these these AI apps, and and we're excited to see where that goes."
"Yeah, I love that, and that's why I was really excited to talk to you. Um, but like let's talk a little bit more about how you built Cove so far. So you've been building it for like a over a year now, right?"
"Yeah. So so how did you go from like, you know, I'm sure leading like, you know, a couple dozen or couple hundred PMs at Uber to building this uh this AI thing? And also and also like what what kind of user problems are you trying to solve at the end of the day with this product, you know?"
"Yeah. Yeah, for sure. So so I can give you a little bit of the context of of the kind of journey of how how we got here. So when we left Uber, there was a group of us that had gone through that journey together, and we were really excited to do something else together. And it just so coincided with an incredible time to be able to explore all these interesting things happening with AI. And so at first we were really like first trying to get an understanding of the core capabilities of AI, and we're building all these like, you know, interesting sort of like prototypes. We built something in gender of music. We built something that was more of an AI travel planner. And as we were doing that, we kept on kind of coming back to almost this kind of core realization, which is those experiences that we're building were very chat focused. And as we started to think about the the capabilities of AI increasing over time, you know, the question that we had for ourselves was, 'Is this actually fundamentally the right way to work with AI?' Right? In a world where AI capabilities can be more analogous to what I expect with collaborating with another human, then how do I collaborate with other humans? And should that also be more similar to how I work with AI? And so that was the real initial inspiration, right? And if you think about, you know, how we might work together on something, then you compare that to a chat thread, then I think you start to realize just like how fundamentally limiting a chat thread is, right?"
"Yeah, that's a really good point, because you know you and I would probably use a whiteboard or like get in a room."
"Exactly. Exactly. When you think about, you know, how people think, first of all, thinking is a pretty messy process, right? You're going to like you branch in a lot of different directions. Like it's nonlinear. Like you explore in a lot of different things. The linear rigidity of a chat thread is really hard to be able to kind of bridge that sort of gap, right? And then to your point, when you work with other people, you need to have a shared context. Uh, and so if we're in the same room, we might be in front of a whiteboard, for example. And so context defined in a chat thread is very hard to be able to do. Um, and so so you know those were the kind of starting things that got us really excited about building Cove. In many ways, we kind of think of it as kind of looking at what has happened in the past as we've seen fundamental technology change and also trying to apply that to AI. And so you know when you look at personal computing moving from command line interfaces to graphic user interfaces, when you look at phones before and after they had touch screens, each kind of fundamental change required almost like a new UI paradigm to kind of get the most out of that underlying technology. And so we think it's very similar in front in terms of AI and chat threads. And so we're just on that journey of trying to figure out what is that right that right interface."
"Yeah, that that's a really good—I didn't think about it that way. I I think we're pretty old, right? So we were around when there was like DOSs, and then you had to put everything into the command line, and then there was like Windows 2 or like Mac where you can have a visual interface."
"That's right. That's a really good point, because now because I think yeah like AI is basically at the DOS stage, right?"
"That's right. Yeah, yeah, yeah."
"And so for each of these kind of like um shifts, there tend to be new primitives that get created, new interactions that get created that unlock more of those capabilities. And so you know that's that's in many ways, you know, what we're really trying to tease out as we're building Cove: like what are those new primitives, what are those new interactions to really allow us to, you know, match what the capabilities of AI are going to be as they continue to advance."
"So how do you uh do this in practice, right? Because like this tech is like advancing pretty fast, and then you're trying to do something pretty groundbreaking, but then you have to use the existing whatever chat competition API or whatever API is it? Is it kind of a pain to do this?"
"Or yeah, it is. I mean, um, well, I I wouldn't use the word pain. I think we're having we're having a lot of fun trying to figure out, you know, how how to how to how to piece this stuff together. But I do think—well, maybe first I can give some context over how Cove is built, um, and then and then talk a little bit more about just how we kind of deal with that challenge you kind of called out. So we feel very fortunate because we can sit on top of all the incredible things that all the other companies build from a model standpoint. And so like as we kind of think about how to integrate those things into experience for Cove is always trying to think through that sort of kind of tradeoff for a particular task: what's the performance you need, what's the cost, what's the speed, right? It's basically kind of looking at all those different trade-offs and then being able to figure out what the right model is to use. And so when you use Cove today, maybe you'll have suggestions powered by Llama, right? But then coding powered by Sonnet, and then Perplexity providing some web search results. And so in some ways, we feel really fortunate because we're able to really tap into each of these things in a way that we think makes sense and put them together into like one product experience. Your the problem that you call out is still a very important one, right? Because you know I think fundamentally one of the just the fundamental challenges of all of us building an AI right now is that as product builders, we're trying to kind of predict the future, and that's already hard, but the future feels like it's changing faster and faster now, right? And so because of that, like it requires kind of like, you know, like some additional sort of almost like skill skills and ways of operating to be able to deal with that. And so you know as an example, as a team, I feel like we have—you know, as a startup, you pride yourself with being able to be really nimble and very adaptive compared to a larger company, whatever the case is. But even as a small startup, we're eight folks—the speed of how AI is advancing also stress tests how we operate as a group too. And so I think you really have to embrace that as you're as you're kind of working through things. I also think that it really causes you to have to be thoughtful about how you focus your time in terms of what problems you're trying to solve. And in addition to that, um, when you think it's it's when you think it's the right moment to release different features. So maybe just kind of like just double clicking across both of those things in terms of focusing on the right problem with this kind of rapidly changing foundation that we're trying to build on, you can sometimes fall into a trap where you might not be focused on the right thing because that thing will actually be solved by a model like in the future. Like we were there—there's when we were building one of our early prototypes, this was before we started building Cove. I remember we were having a chat with someone, very smart person, we're trying to grapple with this context window problem, and he made this comment a conversation where he's like, 'Well, you could spend the next couple months trying to, you know, do some things to try to help with this context window, but I'm pretty sure the models are going to be better at this in a couple months anyways, so you should focus on something else.' And I always remember that conversation as we're trying to work through these sort of things because um like you have to like really um try to figure out like where that focus is. And then in terms of, you know, the readiness of features, sometimes you get the timing wrong, right? Like you might anticipate that a model is going to be really ready for this particular capability, but then it's not completely there yet, and you're like, 'But I want to get user feedback,' and like, you know, you kind of put it out there hoping that it's going to continue to get better. Like, you know, for example, with the generative app, things that we're showing, do I wish that it could code those apps faster? Absolutely. Um, do I think that those that capability is going to happen as the models advance? Yes, for sure. And so we're going to put it out there knowing that that experience is going to improve. We're not going to hold it because um because we feel like you know that's that's that's the way that the the technology is going to be evolving regardless, and so we can embrace that as we build product."
"Got it. Yeah, yeah. I mean, at least we've sit now you can actually build apps that actually work for the most part. So but but I I I noticed I mean it's like a very purposeful UI design, right? Like you said, you're you actually set the expectation that you have to wait because it's not it doesn't just generate right away. You have to hit a button for it to generate."
"So that's right. But but what's an example of something that uh doesn't depend on like the model advancing? Is it like some non-AI thing? Does that making the UI beautiful, or what is that?"
"Well, yeah, I mean, I do think that um the you have these sort of capabilities of the AI, and then you also have the fundamental features of the workspace. And so in many ways, what we're trying to do is we're trying to advance across each of those things. Sometimes those things are coupled, but sometimes they aren't, right? So like the it's not like um the roadmap is completely bound just to the sort of like capability of the AI itself, right? It's like um like there's some features that certainly feel like that's the case, but like I think it's the more interesting questions ends up starting to become like, 'Okay, well, how how much do you lean towards those AI features versus there's other things as well that kind of create the right holistic user experience?' And you know I always ask product builders and founders like, 'Do you have uh you probably have a vision, right? Like of where you want the code to be, but how long is your road map at this point? Like is it like a month or—'"
"Yeah. Well, yeah, well, um, yeah, it's an interesting question, and I do think that you use two different words there: vision versus roadmap, right? And I do think that like it's important to separate out those two things very fundamentally. I think that you can have conviction on a longer-term vision but have a lot of near-term uncertainty of what your roadmap is. And so I think that that's you know maybe maybe an interesting way to describe what it feels like to build an AI right now, right? Is that like you know we all feel like there's a particular future that we're trying to fast forward, and you know there's a reason why we all took the leap to try to like build that, but exactly how we get there, that that that ground is shifting, and so we have to be like really nimble to respond to that as as uh you know advancements continue to come out. So and and do you think is is your vision kind of like that AI brainstorming partner or like kind of like the like do you have a Python statement for this thing?"
"Yeah, well, I think like um the I'm I'm going to use more in this oneline statement because I think that like you know there's there's there's there's there's more that we can kind of like talk about there, but but I I do—for us it's very much trying to think through what is that magical workspace that allows you to get the most out of AI. And so the um when we first started, it was very much around, 'Okay, we need to have this workspace as shared context. We also need to be able to create an AI that has more of a collaborative sort of features that allow it to progress from being more of like a you know intern helper to be more of a fundamental thought partner.' And so how do you create those sort of features? I think you know more recently as we've been building out these features like generative um generative apps, um the the kind of new sort of like dimension to that is, 'Well, the workspace itself can actually be more fundamentally generative.' And so that's something that we get like we find like really interesting in terms of like, 'Okay, well, maybe the right way to collaborate with AI is also like like the workspace itself is actually transforming based on what you need.' And so that feels feels like you know like another critical piece in terms of kind of rethinking what the what that next kind of paradigm is to get to most of AI."
"Amazing. And and you talk about you know you don't you don't want to build what the model will deliver next. So do you have any predictions for what the model—like we're talking about like uh this like a agent thing or you know—"
"Yeah. Yeah. I mean, I think like um we'll we'll if you if you look at um that might be its own podcast uh in terms of to talk about that in more detail, but like um but if you if you look at what has been happening recently, there certainly is like a lot of talk on agent—agent's funny, right? Because I think it's become one of those words where like you might get 10 people and they have 10 different definitions of exactly what an agent is. But like certainly certainly you have you know that as an area—um you have more of like deep research as as something that's like really fascinating as well. And so when you look at those kind of capabilities—let's call agent basically being able to do AI doing more on your behalf, right?—and then let's think of like reasoning as instead of trying to create an AI experience where I am going to try to get immediiacy in terms of the output, I'm actually giving AI more time to be more thorough before it comes back to me. Like those are things that certainly make sense. And I think for us, we kind of think of it as like, once again, if we're trying to think about AI as a great collaborator, are those collaborative relationships that you're used to as you work with other people? Absolutely. I'm used to being able to like offload some tasks to someone and have them work on and come back to me, right? Um, but like when I work with those people, I still need to have a place where I'm orchestrating those actions, right? I'm having those conversations, or the cases I still need to have a place where even when they come back to me, we're able to continue to iterate on that work in some sort of way. And so those are places where we think it actually continues to fit very naturally in terms of having a workspace like Cove. I'm working through a particular problem. As I make progress working through that problem, now I have a way to actually orchestrate agents to be able to do things. Maybe there's particular status that those agents need to like kind of come back to me to like and those based on what they've done, then further things will change in my workspace. You know, or like, 'Hey, I've paused on my workspace, but like I get an email because actually Cove has actually done more work on my behalf and then has told me like, "Hey, you should come back now because now you can actually do these things too," right? And so like those are things I think—in terms of your question—th those those expectations from a product standpoint, I think we'll we'll start to become more integrated with AI products as the capabilities improve. Um, and we very much see that like um uh in terms of fitting well in terms of what we're building at Cove."
"Yeah, I think uh right now, you know, as a PM, I work on I work in like a Slack channel for a project and maybe like some Google doc or like some Figma file, and um yeah, if you guys can—I mean, I'm not sure if targeted the work use case, but if you can—there must be a better way to do that than like endless Slack pins and stuff, you know? Must be a better way."
"Yeah, especially if when you have like multiple AI agents contributing to the project too, right?"
"Yeah, yeah. It's almost like what is your uh what's your inbox for the agents doing work on your behalf, right? And so so when we talk about orchestration, like I think that's like you know similar to that, right? Um, you need a place to be doing that, and so it'll be interesting to see kind of like how those product experiences evolve. Um, so so let's kind of wrap up by talking about uh briefly about your Uber experience. Um, so I I I guess uh you know you built this awesome bit business at Uber, um, and I I guess what I'm getting at is like what were some key skills do you think that the product folks on the team uh had to actually deliver this $25 billion business? Yeah, over five years or something. And then and then the next question I'm going to ask is like what kind of skills did they have to learn this AI era, but let's let's focus on the Uber Eats first."
"Yeah, for sure. Yeah. So so let's talk about Uber. Maybe to give a little bit of context, so I joined Uber in 2014, and it was to help create a new business unit to figure out the new businesses at Uber. At the time, Uber was just rides. And so um the team name was the Uber everything team because we're trying to figure out everything else to do besides moving people around. And so you know the um one of the really interesting things about that experience is actually um how we are able to build that business within Uber. And when you think about large companies who have been able to successfully create kind of like that next line of business um at that at that magnitude in terms of size, there's not really a huge list, right? I mean, Amazon AWS definitely is like you know one of the like like really good examples, but it tends to be less common for these things to happen. And you know a lot of times like—"
When I kind of reflect upon our time at Uber Eats, it's trying to like almost think through, like, what are the kind of right conditions that you need to be able to kind of have the right internal innovation to be able to create that next line of business so successfully? And I think for us, there was a couple things that just helped—was really helpful for that almost initial formula when we formed the team. Um, the first one, and you know, anytime you talk with people that go through just like, um, like a crazy building experience, they're probably going to call out the people, right? And absolutely, like that's that's the first one for us. But I think the specific thing I call out from a people standpoint is the initial group that worked on Eats was a really interesting combination of some folks that had been at Uber for a very long period of time—some of the original ops people, some original engineers—and then and then they were mixed with a new set of folks on, for example, the product and design side. That mixture of new perspective, but also folks that really knew the Uber culture and really knew how to get things done at Uber, was a really interesting kind of like cocktail for an initial team, so that that's the first thing I call out.
The second thing, interestingly, I guess I would call it almost ambition. Like, um, when you're when you're when you're a startup, you're almost like just always trying to fight for survival, right? The next, you know, being able to survive the next day is like, you know, success, right? And you want to keep on going. Interestingly, when you're trying to do something within a larger company, you have the shadow of the other business always, you know, casting its like it's it's it's its shadow on you. And and so for us, you know, when we set up the team, the the way that we talked about it was our goal is to build a business that's the same order of magnitude as the rides business. And it was both a very ambitious statement, but it was also a very clarifying statement. Because I think that whenever we iterate on a lot of different ideas, and there are a lot of moments where when we were working on something, I had some level of success, I was like, "Oh, you know, if we were a startup, we'd be really happy over how things are going right now." However, do we actually believe that this idea is going to be at the same size as rides? If so, we should keep on going; if not, it's time for us to continue to pivot, to continue to iterate. And so being able to like really have that sort of anchor of expectation and be able to like have the paranoia of like, "Okay, actually, is this is this going to be the right thing," causes, I think, to get to a better um, a better results and a better a better business.
Um, and then the third thing would be: when you're doing one of these things, a lot of times, like when companies are trying to think about creating like another line of business, sometimes they structure it as almost like a hedge bet, right? It's like, "Oh, we should have a group of folks experiment on this particular thing because, you know, maybe this could be like an interesting thing for for the business." The problem is, if that is kind of your mindset, then whenever things get hard, then you start to deprioritize those things, right? The core business that's working needs more people to do XYZ; okay, let's have less people work on that experiment. And then those those those um, those those new ideas die. And so for us, it was very different. I think, like, for Uber, everyone collectively agreed, from the leadership all the way down, that we needed to figure out these new businesses because it was existential for Uber as a company. And because of that, I think it caused, interestingly, more of almost like a patience to be able to like take the time necessary to actually figure out what we needed to do. So we actually, I think, had the right kind of cover and executive like buy-in to be able to actually like go through all those different pivots before we actually figured out what was Eats and be able to like grow out that business.
Did you keep the team relatively small, too? Or the team in the beginning was quite small? Yeah, it was it was we had a small set of folks that were, um, you know, a bunch of like really strong generalists, being able to like just do a bunch of really rapid experimentation. Um, and so we kept it in that sort of mode for quite a while before before we found the thing we're like, "Okay, now it's time to scale," and then we went through this really crazy sort of growth period. Yeah, you got to be like a VC kind of like, you know, with the founder—just leave the founder alone. Like, you can't be like, you know, "What's the OKR every quarter?" That's right. That's right. It's exactly right. And so, you know, the the like you you need to have leadership that truly are founder in their perspective to be able to also have an internal team be able to operate in that particular manner, and that's what we found, and that that was really really important. Awesome. So I think this uh whole like team of generalists thing is is like really is kind of like a hopefully a trend with this uh now that we're entering AI and and um, you know, I I feel like a lot of PMs, you know, like are like, you know, big companies like kind of like writing docs all theirs doing something or trying to align the CEO. Yeah, and um, how can you any advice for for them to have a better life? Or I I I I I do I think agree with the the sentiment that as organizations have grown, um, you've seen more of this sort of specialization with, you know, various functions, but PM being one of those, um, and it was, you know, sometimes as you're scaling out a business, like that's the type of team that you need to build, right? But the um, but the world's also changing, um, and like as you start to see like more individual empowerment with all these AI tools, then I do think that the trend line is going to be reverting back towards having people that can be generalists because they're able to get the leverage with these tools to do a lot of different things. Um, so so we'll we'll see we'll see if that's the case, but I guess, you know, we're we're now on on record of at least, you know, predicting that that might be the case.
The um, in terms, I think one of the questions you asked was was how PMs can, you know, prepare in terms of almost like their AI sort of skills, right? Be successful in AI. Um, and there's probably two things I would—two kind of categories maybe I call out. The first is just like in terms of almost like how you operate as a PM, um, and you know, there's a lot of things that we can talk about around like how you can use AI to be more efficient as a PM, um, but maybe the thing that I would call it, which is a little different from that, is we're should be progressing to a place where people almost kind of rethink how what their fundamental relationship is with AI, right? If like it's really easy to humanize AI, right? So you can think of AI as like, you know, the intern that like helps you do research on the side, right? But that's very different from thinking of AI as a true collaborative thought partner, right? We um, on the Cove team, we have this this picture of John Lennon and Paul McCartney singing together at a piano, like doing like making music together, and we almost like we love that picture because we hope that one day we can have that sort of relationship with AI as well, right? How can you actually get to a point where you have truly a collaborative thought partner that can actually work with you in that sort of deep manner to bring out the best in terms of what you're trying to do? And I think that like, you know, there's already the capabilities of the models and also the AI products that people are creating are allowing you to try to work with AI in that sort of manner, right? When engineers are wanting to like like, you know, make progress on something, they pair together on things, right? Like when you're working through a problem, you want to be able to get like a thought partner, and with AI, hopefully now everyone has a thought partner available to them anytime they want to be able to bounce ideas off someone, right? And so as a PM, I think you want to start to be able to try to work with AI in that particular manner.
Um, and then the other the other category I call out is is actually just like being able to just like jump in and get more fingertippy in terms of your intuition with AI. I feel like there's a lot of PMs I have conversations with where they may not be working on something on a day-to-day basis that's directly related to AI, and so as a result, they haven't really taken the time to like really dive in and really understand and learn um and tinker. And I think a lot of times it's actually—I feel like I' I've heard almost two different things: one is like sometimes PMs overintellectualize, so they're like, "Oh, like I need to find a class to take so I can learn about transformers or like read the transformer paper," is almost like they feel like they need to kind of like, you know, go through that sort of almost like more academic route. And then sometimes you have PMs—I don't know—they like they'll listen to a podcast where the cases and they'll hear like, "Oh, PM skills is going to be really important to write like evals," right? And they're like, "Oh, like okay, I'm going to need to learn about that particular thing." I actually think it's like a lot simpler than that, right? Like I think people just need to start—like you need to like be able to break out from your day-to-day working on whatever you do to just like get—start to tinker and get more fingertippy in terms of playing with these different products because you like you start to build more of a fundamental intuition over what AI can and can't do. Um, on the team, we almost kind of call it like your AI spidey sense, right? You start to kind of almost like start to get that sort of understanding almost on a per-model basis, right? You start to kind of do more things. Um, and if you're able to build that sort of like kind of context, then you can start to connect the dots over like how you're going to apply AI from a product standpoint. And it's easy to jump into, but I just feel like there's a lot of people that just like feel like for some reason they're not doing it yet, um, and like like um, and so that's what I really encourage people to do. So yeah, it feels like a lot of people don't have the agency—like they need someone to walk them through this stuff, like, you know, "You got to learn to course or like, you know, do other stuff," but like with AI, just jump in—just jump in and make a bunch of mistakes. Like, you know, "I've been vibe coding and I don't know the hell I'm doing, but like it's fun," you know. Exactly. Yeah, I mean, you can you can go to an AI and you can almost just like approximate a backend for some service by just creating some sort of prompt and then just see how the responses go, right? I mean, like that's even without all the prototyping and vibe coding other things that like, you know, are starting to kind of emerge, so. Um, so yes, so the barriers were already low and are continuing to drop in terms of people being able to like jump in and get that understanding, so so I think that's the first step. Awesome. So that's a pretty good message to end on. Yeah, like I I personally probably talk to AI more than talk to my co-workers at this point—hopefully less than my wife—but but uh, yeah. And also like uh, make time to tinker—I think that's really important—that that is like you don't need like these frameworks for product sense or whatever the hell—you just need to actually use the products, and that's how you build a process. It's like, cool. Well, um, where can people find Cove? Yeah, so you can go to cove.ai. Um, it's free to try, so definitely encourage you to um, to come check us out. We love getting your feedback, so we read every single piece of feedback that comes in; we try to talk with as many users as possible, so we'd love to hear from you. All right. All right, thanks so much, Stephen. It's been an awesome conversation. Yeah, thanks. Appreciate it.