Big Technology Podcast - AI Agents: Mirage Or Real Revolution? — With Dmitry Shevelenko

Episode Date: May 7, 2026

Dimitri Shevelenko is the chief business officer of Perplexity. Shevelenko joins Big Technology Podcast to discuss whether the AI industry’s shift toward agentic 'super apps' and computer-using assistants will become a real business. Tune in to hear why Perplexity believes computer use is a durable productivity tool, how it thinks about consumer AI’s growth slowdown, and why multi-model orchestration could be a competitive advantage. We also cover AI search, enterprise adoption, trust and permissions, Chinese open-source models, pricing, and the economics of AI demand. Hit play for a sharp conversation on where AI products are heading next. Join Big Technology's AI Summit In San Francisco: https://summit.bigtechnology.com/ ——— Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice. Want a discount for Big Technology on Substack + Discord? Here’s 25% off for the first year: https://www.bigtechnology.com/subscribe?coupon=0843016b Learn more about your ad choices. Visit megaphone.fm/adchoices

Is the near uniform move of AI companies to agent super apps going to pay off? Let's ask Perplexity's chief business officer right after this. This week, I'm live at Knowledge 2026, ServiceNow's annual conference in Las Vegas, where Enterprise AI moves from promise to production. I'm sitting down with ServiceNow's president and CPO Amit Severi on the platform strategy powering at all, their people and technology leaders on what AI means for the workforce, the engineering team behind ServiceNow's Nvidia partnership, on what it really takes to ship AI at scale and Ultra Beauty on deploying AI across 1,300 stores.

These are the conversations you won't hear anywhere else, and new episodes are dropping this week on my YouTube page. We've all heard the stat. 95% of AI initiatives fail. It's not because the technology isn't ready. It's because you don't have the right process or the right partner. Meet Aboard. A board is your partner for AI transformation, which means they listen, use their very own

powerful software tools, and deliver exactly what. what your company needs to thrive in the age of AI. Working with big and small clients, A board always delivers in weeks, not months. Your AI revolution is just beginning. Visit Aboard.com to get your AI rollout right. Welcome to Big Technology Podcast, a show for cool-headed, and nuanced conversation of the tech world and beyond.

We have the chief business officer of Perplexity here with us. Dmitri Shevillenko is here with us in studio, and Perplexity, as you may know, is one of the many companies moving towards this, super app style product with perplexity computer. Now they are joining OpenAI with Codex, an Anthropic with Claude Code, as one of the many companies moving towards this agent

that can control your computer and get stuff done for you. And today we'll talk about where that's going and whether it's going to be a real business. Dimitri, great to see you. Welcome to the show. Thanks for having me. Looking forward to the conversation. So we're here in mid-20206 and I gotta be honest.

I thought at this point you would be a subsidiary of Apple hasn't happened yet. Well, sorry, you're probably market, bet there, you know, didn't pan out. Just to be clear, there was no polymarket bet. I just thought it was a good idea, but it hasn't happened. You know, we have a great blossoming partnership with Apple. They actually are really excited about what we're doing with personal computer and how it uses

Mac minis. So it's a nice growth area for them. Yeah. So that is, you know, we found a way to work together there. But we're having too much fun being independent. and a lot of the world is realizing that the power of multi-model orchestration, mass multi-model orchestration, what was first a rapper is now a harness. So we're really excited about the future ahead.

Yeah, and that's, to me, the main criticism, I was obviously very vocal saying Apple should buy perplexity. I think they actually gave you a call. I'm not taking credit, but maybe I contributed. The reason why I thought it would be a good tie-up is because you know, all the criticism was, oh, perplexity is just a wrapper company. And I was like, these guys actually know how to build AI products. Obviously, the search engine, the browser comment looked pretty cool. And then this new computer application where perplexity will

take over your computer on your request and do things for you is really where AI is heading. And as you mentioned, it accesses multiple models as opposed to just being tied to one. So I thought that would be a good acquisition for Apple, which has clearly struggled to take these models and translate them into working products, at least so far. Maybe they'll figure it out with Gemini. What do you think about their CEO, John Turnus? Or their incoming CEO, John Turneris. Well, Apple's always been an incredible hardware company. And I think this is, you know, an era where hardware will matter even more because software is going to face waves of companies. commoditization pressure.

So I actually think it's a really smart pick, and we're excited to see what they build. And we want to build really powerful solutions that work well with Apple hardware. Okay. We're going to get – you have a partnership with Samsung. So we'll get to that in a bit. Let's not bury the lead here, though, which is that, you know, perplexity gained, I would say, mass awareness, at least in the tech industry, because of the search engine that you built.

Arvind, the perplexity CEO, was very vocal in saying, we're going to take on Google, we have this new way of doing search, and look out. And when we look at the usage of consumer AI, something very interesting has happened over the past, I would say, six months, which is that use is pretty much flatlined. If you look at the DAUs of generative AI apps, from, you know, Apopia, for instance, there is sort of a flattening that starts late 2025. Even looking at perplexities market share of AI search, it was close to 20%. I think this is, again, according to Aptopia, mid-20205, and it really has decreased,

kind of flat over the past month or so, according to similar web, your traffic, about 5.2. million average daily visits, up 2% over the past month, compared to 182 million for chat chipped, which also isn't growing too significantly. That's up 5%. The question for you is, everyone is now pivoting to this super app, this app that can control your computer, you guys, opening eye, I'm wondering, is this happening from a position of strength, which is that, okay, we're just going to move here because the technology is so strong? Or is it potentially a reaction to the fact that consumer AI hit a ceiling and you need something else? So, well, I'll tell you that I don't know those metrics that you shared, but the stats I look at every morning is our revenue.

And we started the year at under 250 million error. And Aerevan recently shared that, you know, as of a month ago, we crossed 500 million error. And so clearly we're creating value for our users. And when we actually go back and understand who was using perplexity, even when it was more focus on, let's say, consumer AI as you define it, people are actually using perplexity for work and knowledge-related tasks. So they were coming to us, you know, as much as we were talking up, you know, this is the Google search killer.

people were using perplexity to get ahead at work. Even when they weren't using the enterprise version, this was their secret weapon to be more productive, have greater leverage as they build businesses, create businesses. And so in some ways, we're not, we haven't shifted our focus. We're really going to meeting our users where they always were. And what's possible now is, and this really started, you know, you couldn't have built something like computer before November, December of last year. Because model capabilities advance where you can have longer time horizons for running tasks, right, where you're not just answering a question, but you're actually doing work as an agent on behalf of the user. And one thing perplexity is always pride ourselves on is being the best at understanding what the new emergent capabilities are and finding ways to make that accessible and useful for a broader population. And, you know, that's where we focus.

But I think revenue is a much more honest metric than kind of, you know, top line MAUs, which I think, you know, can include in. a lot of hype and exploratory activity, but aren't as tightly coupled with value. Okay, but I'm going to give the alternative perspective here, which is that the MAUs matter. Like typically, MAU, of course, monthly active user. When you're typically in a growth surge, you start talking. I mean, every company, every tech company, they grow users, and then they have this big user base, and then when the growth slows, you start hearing about average revenue per user. You need more users to have a bigger user, to have a bigger revenue base, don't you?

Well, we're not talking about average revenue. We're talking about total revenue. Right, totally revenue. So pay, we're talking about it. Yeah, I guess that's the next step. Yeah. I mean, I would say historically, that's been true for consumer internet companies because MAU is a proxy for ad revenue, right?

And as has been reported, like, we're not focused on advertising-based monetization. we realize that there is when a core value prop of perplexity is accuracy, is really hard to reinforce that to users when you also have ads running alongside the answer. And so I think some of why MAU matters less is, at least for us, is we're not trying to go to advertisers and say, look at all these users that you can show ads to, across all these different demographics. So that might be part of the shift in focus as well.

Yeah, I mean, you could, you could, the, to support your argument, Anthropic does not have the lead in users whatsoever and doing crazy amounts of revenue. So if you figure out this enterprise use case, you could be a massive company. I mean, we're looking at, they're both, Anthropic and opening I are both going to have trillion dollar IPOs, and we'll have many large companies, I think, that will follow them in the generative AI world. but let me get your take on on if you well let me just get your take on the consumer side of things and then we'll move more on the enterprise side I mean even if people are using these products for work they're such powerful tools and you know they were like chat chTPT was the fastest growing consumer product ever I guess it still is but that growth has tailed off what do you think is behind this

flattening of consumer AI product growth overall. Let's just take it with the whole industry because it's certainly happening. Is it just that like they kind of hit saturation or is it, you know, we know there are fears about AI? Is it people are just too afraid of AI? What's your best diagnosis there? I think there's some of the use cases got ahead of where people were curious to explore like what is this AI thing, but their behaviors didn't change. But I also think there's a fusion of consumer and prosumer that we find very interesting.

A lot of people are now empowered to explore launching a side business or, you know, explore, like doing that, you know, that project that they never had the activation energy for. And now because you have these super powerful tools at your disposal, you're more than happy to spend, you know, money behind that because you feel like you get leverage there. So I think consumer to us is not just people using perplexity to look up the weather, right? You don't need AI for that. And so I think part of what the broader industry needs to do is educate users on what is possible now. people refer to this as the capabilities overhang, right? Where the models got a lot more powerful, especially in the last six months,

and people are still using them in a very, you know, web 1.0 way. And that's just going to take time for that discovery to catch up. But we're, you know, I'd say this is less relevant for perplexity, but I'm confident that everyone will prefer to have a more intelligent set of software. to help run their life. Webpoint 1.0 meaning like information retrieval. Yeah, just like the most basic. Yeah.

Yeah. Like, okay, like sports scores, you know, like weather, you know, basic news. Like that's, you know, that's where still a lot of people are. You don't necessarily need, you know, these new agentic capabilities for that. There's all kinds of, you know, other things people can be doing. And the thing that we're going to realize is the constraint on making the most of AI is our own curiosity, right? Like, you know, that's the bottleneck.

And that's why, you know, perplexities, you know, we design our products to spark curiosity, to activate it to, you know, that's a big part of our brand is curiosity. Because like when we, when we kind of zero out, like, you know, what gets commoditize, what doesn't, the uniquely human ingredient to taking advantage of all this will be curiosity and agency. Let me give you my belief on why we're seeing this slowdown. And we can sort of, because this does lead right into the agentic use cases. When we've seen the biggest spikes, they've been around some of these multimodal use cases. So not text. I mean, chat chit got to 200 million users because of text. People were interested to see what AI could do.

So I think that novelty and that interest, you know, built the foundation. but where we were we were called she's opening i for an example where opening i saw the biggest surges was after voice hit remember that demo where it sounded so much like scarlet johansen she threatened to sue open ai you see an inflection point and growth there and then images the studio jibbley moment still was just one of the like i need i mean i know somebody that created like seven open ai accounts just to because they kept getting rate limited on the usage yeah and so of course you'll probably see a user spike there even if it's not, you know, individual users. So that to me is like as that, as companies have shifted away from those things, we know that SORO is going away at OpenAI.

Obviously, they're still doing images. They just released a great second generation of their latest image product, Open AI did. But there is going to be this sort of moment of adjustment among people from going from what the AI companies were initially telling them, you know, chat and images and voice to this new use case, which is like, we think that the model should take your computer over or whatever, the model through a harness should take your computer over and let you do stuff, and that will naturally lead to a divot. Yeah, I mean, I think I agree with the thesis, right? A lot of those spikes in usage were novelty driven, right? Like, I mean, your friend that created the seven opening eye accounts, you know, I bet they haven't created any studio Ghibli images in the last 30

days, right? Like, I don't see those around anymore. It's probably gone from the family chat. Yeah, yeah, it is. Though you still see some people's profile pictures are like Studio Ghibli. And so that is a warm reminder of that era of AI. I think the novelty spikes are great because it raises, you know, broad awareness. And it brings, you know, it brings people in.

And then people have to, you know, discover their own kind of habitual use cases. But you can't, yeah, you know, novelty is what it is. I mean, nanobanana had a similar, you know, moment for Gemini. And I think you could see now it's kind of, you know, there's been a reduction there too. Ultimately, like, we see value in the most economically productive aspects of AI, right? And that's why, you know, for us, a core foundational investment has been accuracy. And you almost think of search and accuracy is, you know, two sides of the same coin, right? You need to have best in class search so that whatever you're doing with AI is grounded in the most up-to-date, you know, highest quality sources, best snippets of that information working for you.

And so I do think the, you know, I don't think it's fair to. call us what we're doing a pivot, but I think we're mapping our investments towards what are the most economically productive uses of AI that have the most enduring value. And effectively what's happened now, and I mean, you're probably a great example of this. You know, you're running, you know, an independent business, right? That previously, if you were not using AI, which I'm sure you're using in many big and small ways, you'd probably need to hire, you know, a lot of people, marketing agency, maybe a software

developer. Yeah. It is crazy at being so heavily invested in learning the tools, what you can do with that. Yeah. So, like, I mean, you're like the, you know, we should do a case study on you because you're, you're exactly like what we see is the future of the economy, right? Like someone with high agency, right? You had a vision of, you know, running your own media business that, you know, hopefully

one day becomes a media empire. and you're able to make very quick rapid progress on it because you have a team. You know, I think of it like we all just got 100 employees, right? And the shift we're seeing in both prosumers and in the workforce is everyone now gets to operate as an executive because your job is to wake up in the morning and think about, okay, what are the useful tasks that I can, you know, deploy the 100 agents? that are on standby to grow this thing. And so that's a, that's very, again, very different than like, you know, casual chat and generating images.

Like I think those things feed into each other because sometimes, you know, the spark of curiosity requires kind of the quick question and answer. And so you want to make that minimally, you know, you want to make that delightful, easy, low friction. So then people are inspired to go after the longer horizon tasks. and so we see them working well together. Right. But, you know, the future of AI is what you're doing. Yeah, and it is interesting because I do use these.

You know, I just cited the groups I wouldn't need to hire because I'm using this stuff well. But by having access to the tools, I'm actually able to do a lot more, I would say, economically productive activity than I would have been if I wasn't constrained by them. So, for instance, because I'll have like a little extra margin because I don't have that marketing agency, well, maybe I can use that to host an event. Right. By the way, folks, we're going to be doing on June 18th, Arvin Srinivas, CEO of Perplexity is going to come speak with us.

I'll link it in the show notes. If there are still tickets, you should definitely join. But that's something that exists because, you know, there's a little bit higher margin, and we can invest in doing an event because of that. So I think there's like, we'll see a very interesting transformation of the economy of this stuff works the way that many anticipate that it will.

and I've never really been bought into the gloom and doom hypothesis around it. But I guess that's a different discussion. Let me just sort of ask the natural follow-up to what you just said, though, which is if chat, images, voice, were part novelty to cause this explosion of interest in generative AI. Why are you sure that this computer style use or super agent use, case is not going to be similar. For instance, just to make the bear case, maybe it is also a lot of people trying out these apps and saying, oh, that might be useful, but then there could be a pullback from, I'll just give one example, and I'll turn over to you. Think of my teeth into

perplexity computer, which is perplexity's agent or super agent, I guess, is the best way to describe it. And I added suggestion created a daily digest email for myself. So it connected to my Gmail. It's connected to my calendar. It tells me which emails I need to respond to. What's going on today. What I should be thinking of. The headlines, it's pretty cool.

But is there also a chance that that could just potentially be like, oh, that was kind of a cool new use case, but not like a revolutionary use case? Because you could have said the same thing about chat images voice, that they were cool use cases, potentially revolutionary. Maybe they're not. Maybe they have potential to be that way. So why is this not, you know, another one of those novelty use cases? Yeah.

So what we're seeing with computer is people are generally using it the way you were describing the way you're running your business where it's like you now don't need to hire, you know, dedicated staff or a dedicated, you know, agency to do your marketing, to do event production. you're gaining leverage from these tools, right? And what we're seeing is the longer people have had access to computer, I mean, this stuff is still brand new, but they're using it consuming more computer credits every week than the previous week, right? So we're actually just in the extreme upward part of the ramp.

That's a big part of why revenue is ramping as well. So we're certainly not seeing that. And I think the fact that, I were people are now meant the mental model is not this is like I'm spending on software. People are thinking about this as, you know, this is actually part of my payroll budget, right? I have a team of digital agents, digital workers. And, you know, sure, like, would the workers have to, like, show up and do a good job to earn their paycheck? Just like, you know, people do.

but their capabilities are, you know, increasing, and we're getting better every day of connecting the models to different tools, you know, improving, you know, the virtual machine that it runs on. And so I think the nothing, none of the usage of computer right now that we're seeing has a novelty effect. It's all kind of, you know, being tied in, or people are willing to pay for it. is tied into those economically productive scenarios. So we're incredibly bullish on it. And as people in AI like to say, like the models are only going to get better from here, right? So the capabilities will increase. I think consumer is really hard to get right if you don't have network effects.

And so again, I think some of the studio Ghibli, like the voice, those early video gen examples, think that's very different than what we're seeing with computer now. So what should, I mean, you mentioned that people, as they use it, they use more credits. Yeah. What are some of the use cases that you're seeing? I mean, my email, I think is pretty fun. I let that go. But I also see taxes. Yeah. I mean, it's any, so we actually are launching this week 36 different workflows that go on top of computer. So this is everything from building a financial model of a company to filing your taxes. If you're a wealth manager

prepping for a meeting with a client, and again, this takes advantage of connecting to your internal data systems, your snowflake, your data bricks. Just last night, I ran a analysis of what are the models that are being used inside of perplexity right now? What's the distribution of between, you know,

is 4-7 and GPD and Gemini, and it got a very elaborate result back. And I know zero-squel. I can't code if my life depended on it. And I didn't bug a single data scientist at Perplexity. And I was able to do this because we connected Perplexity computer to our snowflake. And I was able to pull in that analysis within a few minutes that in a previous world, You know, that would have been 10 emails, and I certainly would not have been able to get it at midnight as I wanted to kind of dive into that, right? So what we're seeing people do is be able to operate with much greater velocity, whether they're accomplishing marketing objectives, analytical objectives, like building product.

you know, we're now able to prototype new features instantly. We have people on our content team that submit pull requests, basically ship, you know, code that goes live into production without engineers being in the loop. And that's all being run through a perplexity computer. How much can you trust this stuff? You know, again, going back to this taxes example, I don't trust it to do my taxes. Am I just a Luddite? Or is there legitimacy to the worry that if it gets something wrong, I could get a letter from the IRS? Well, actually, I would flip it the other way.

The way people are using computer is to double-check the work done by their accountant and finding significant errors done there. Well, right? So it's actually one of the workflows that we're most excited about is called Final Pass. And you submit PDF, a presentation, a spreadsheet, and it basically does a, a detailed fact check on every assertion and claim in that document and both in terms of fact checking against the outside world and then for internal consistency. And we actually ran through a Gartner press release about their earnings and found like four

glaring, you know, like mistakes in it where they like misstated the earnings. And, you know, we're going to have a fun marketing exercise. basically go through public companies press releases and run final paths through them and show just how much, you know, error lives in the world right now. And so I think, you know, there's – but to get to the heart of your question, I think there's always going to be three fundamentally, like, human activities when it comes to using AI. One is we talked about curiosity, right? you have to give it the spark. Like you have to define, you know, we say, you know, we're shifting from an era of instructions to objectives, right?

So you have to define what are the objectives for, you know, what is the marketing success that you want to see? And then the AI will accomplish it for you. So you need the agency. The second part is just like you need to, you know, error correct and double check the work of a human, we need to get really good. at understanding where AI might go sideways and, you know, do validation testing. And that's going to mean different things in different use cases.

And then the third piece is good taste, right? Only humans are going to deeply know what other humans will find interesting and cool. And I don't think AI is going to – AI can be a great brainstorming partner, but ultimately that's going to require discretion. And so, yeah, I think that – you know, fact-checking, error correction, those are going to be essential skills. But it goes both ways. Like, you know, as I said, you know, with taxes, there's plenty of errors that humans are making right now. And let's use AI to catch those. The question is, if people will stop at,

people will use these tools the way that you intend or whether they will just say, all right, screw it. I'm going to replace my accountant entirely. But I guess you're responsible for that if you do that. Yeah. I mean, just like your responsibility. So if you hire a cheap accountant, you know, and they mess up, like ultimately, you know, that's going to create a headache for you. If you use a bad AI or not using it properly, you know, that's also on you. So, you know, accountability doesn't go, you know, go away with AI. And, yeah, we need to develop a good sense of how do we, you know, like I have a good way of spot testing, you know, when I get an output from AI. like what are the things I'm going to like double click on to make sure there was no silly mistakes.

Yeah. And I love the final pass idea. I mean, I've been doing that for all my stories. I like, will upload the interviews and then upload my draft and be like, where did I miss? What outside context is there that I should be considering? And so it's just natural that type of approach would be applied to other things like taxes, financial projections. Even, I don't know, marketing presentations could be thrown in and be like just triple check the numbers, which I've been doing. And it's quite good at that. Yeah. I mean, the really fun. one was I presented to the senior leadership of Bain, a management consultant, you know, management consultancies, they publish all kinds of, you know, reports. And like, we had a lot of fun, you know, showing them some, some errors in some of the public reports they found. And, like, the people that worked on it were in the room. And so they were, they were giving each other, you know,

some trouble for it. But yeah, there's, I think there's still a lot of value to unlock. using AI to fact-check humans. Okay. But to get this to work right, you have to trust a company like yours tremendously, actually. Let me just read you some of the permissions I had to enable for just my daily email. See and download. I can't believe I actually went through with this, by the way. See and download contact info automatically saved in your other contacts.

See and download your contacts. See the list of Google calendars you're subscribed to. See add and remove Google Calendar. you're subscribed to. View and edit events on all your calendars. View availability in your calendars. See and download any calendar you can access on your Google calendar. Read, compose, and send emails from your Gmail account.

See and download your organization's Google Workspace Directory. I guess I see now why people are working on the Mac Mini. Because, you know, and this is enabled for me right now as we speak. That perplexity has all this access to, like, you know, all of my mission critical. you know, technological infrastructure. I mean, maybe computer right now is like writing up client emails and sending them. I don't know. Well, you do know, right? Because you're ultimately, you know, you're choosing to initiate the tasks. Like nothing is happening kind of autonomously, right? Like, again, the agency is still, you know, human triggered. Like, you're ultimately still

directing. And, you know, you don't need to give all those permissions to get a lot of value. out of perplexity computer. I mean, this is a conversation I have with many businesses is, you know, start with zero connectors and just, you know, see the value there because there's a lot you can do with, you know, just interfacing with all the outside world's data and making more sense of it. But you're ultimately, you know, to unlock the full value, if you think about this as a digital worker, you know, if you hire people, you also give them access to even greater permission right? And people make mistakes too, right?

They tend to work slower than the AI does. Yeah. And, you know, again, another, like, you know, crawl, walk, run that I would suggest is we have the capability for businesses to allow for read access, but not giving right access. Meaning it can, you know, it can create the daily digest, but it won't send the emails on your behalf, right? Which is, like, that's the part where people are like, well, what if it, like, goes and, you know, spams a thousand. folks with, you know, with the wrong. Confidential information. Yeah. So, again, so that's like the read, right.

I think that's like a way, you know, and again, we, you know, with our business versions, we offer very granular controls. And I think that that's the path forward there. But we spend a lot of time getting the engineering on this right. You know, one of our advantages in the space is the only thing we do is build the product. We don't train, pre-trained foundation models, which means all our locus of effort is exactly on making those interactions, you know, first of all, transparent to the user, right? You were able to know exactly what you're giving us permissions for, and then make sure that, you know, it is errorproof in terms of adhering to those permissions. So do you think that the technology today is trustable enough that what I did is not crazy?

And if so, why do you think so many people are running this on a Mac Mini? I mean, there was a Mac Mini in your ad for Perplexity Computer. Oh, so the Mac Mini is actually the other way where it lets you get even more, right? Because with a Mac Mini, you can then get access to your I messages, which you can't with the permissions you got there. With the Mac Mini, also, the agent can run 24-7, right? Even when your laptop is closed, it can run those long horizon tasks. So I wouldn't necessarily interpret the Mac Mini as like, I want, because the inference is not yet happening locally, right?

It's still happening. Do you think it will? Well, I certainly think that as models get more powerful, you will certainly be, and as local CPUs get more powerful as well, you're going to be able to distill powerful reasoning models to a size where they can run on a Mac Mini. Now, I'm not going to offer you like a timeline on, you know, when that's going to, you know, when you're going to get the 80-20 where some of these workflows can shift towards local inference. But I think hybrid compute where certain tasks will run the cloud and certain will run locally. I think that's a pretty safe bet to assume that that will be like the, you know, the right way to anticipate how these systems will work in the near future.

Yeah, that's the bare case to the data center buildout is that eventually you do all the training in these massive data centers and then you sort of distill it and run locally on a Mac Mini. Well, again, I didn't say 100% low. I said hybrid. But like if the work that you're doing in the cloud is so computationally intensive, you might still need all that data center buildout, right?

So I don't, you know, there's kind of, I think we're under anticipating all of the broad broad types of computation that more powerful models will, you know, bring to bear. And so I, you know, from the perplexity point of view, like, we don't have strong opinions on the data center buildout, but there's nothing I see that indicates that that is, you know, a bubble or anything like that. Yeah. Okay.

So just to sort of wrap this part of our discussion, the Mac Mini is not a way to ensconce the agent away. It's to give it access to more and let it work harder. Yeah, and again, with kind of even more granular control, right, and more access to your local files, obviously you're giving those granular permissions, but yeah, we're currently those systems don't support local inference. Obviously, you're doing this. We've just heard at length from OpenAI on this show about their ambitions to build this super app with codex at the heart of it. That obviously will take your computer over.

they call it a new way of using a computer. And then, of course, Anthropic has done this with Cloud Code and Cloud Co-work, which I can't believe how I'm still stunned at how much permission I've given these things, but the payoff is pretty intense in a good way when you do. I guess you've got to take risks in life. Why is perplexity going to be able to compete with these two giant companies in the same product arena? Yeah.

So when we first said about building perplexity, we made a very intentional decision to be model agnostic. And that was kind of very contrarian at the time because the easiest way to raise capital in 2022 was to say you're training a model, you know, especially with our founder's background. That could have been a very easy story for them. they believed back then, and it's proven to be the case, that models would end up specializing. And that is actually one of the most powerful things about computer is on a single given task, it will use different models for different parts of that task, right? So I have little kids, and I love, like, whenever I'm trying to get them to learn about things, I'll create like mini podcast for them.

They're very personalized. And when I do that, computer will use, this is kind of, and this changes week to week, but it'll like to use Opus for planning the task. It will use GPT models for writing the script because GPT is a good writer. It'll then use Gemini models for generating the audio. It will then sometimes actually use great. Rock for fast research, because Rock is a very fast model. It will use Sonnet for writing the Python code to stitch together all the audio clips.

And that's just in one single deliverable task. It used four different models. So the one thing that Codex is never going to be able to support is running Gemini models. You know, it will always be in the GPT family. Same thing for, you know, Claude, like they're not going to, you know, have GPT models. Gemini is not going to have GROC models. So our value as a multi-model orchestrator and being an aggregator is we can tell a user, whatever is the best intelligence that exists in the world today that can help you accomplish your task,

we're going to be using it. And we're not going to be discriminating because of the models we happen to train or the ones we have a special relationship with. And that is a very powerful value prop. And that's something that endures over time. I think the second piece that is foundational that I spoke to briefly earlier is accuracy. When we were focused on the V1 of perplexity, which was ushering in this transition from links to answers, the core technology investment we made in our own tool was search.

you need the most accurate grounding so that whatever the intelligence is processing, the source input is as high quality as it can be. And so that's something where we have a very powerful data flywheel that's been running for over three years of compounding. As people use the product, we see which snippets the models use, which ones they don't. That reinforces the intelligence of the index and what we do in search. And so accuracy is another thing that is very differentiated in perplexity computer compared to some of those other products. And so, you know, and I'd say the third structural differentiator, this one you're going to say might be like soft and fuzzy, but I think it matters is usability. You know, when I talk to businesses, something I, you know, comes up often is the alpha for a company that is not. not an AI company, is not in them building their own internal tools with AI necessarily.

It is in the depth of their adoption, right? Like, how do they culturally, how do they, through training, you know, through the right type of management, actually get everyone to use these superpowers the way you're using them, right? And you're doing it because you have to, right? because, like, you wouldn't, you know, like, you're seeing the necessity. And so, yeah, I'm a psycho who likes to pressure test these things. No, but you're seeing, but it's very useful to me. Yeah, like you wouldn't be, I mean, I don't think your type of business model would work

necessarily with that, with, I mean, it would be much harder. Yeah. It would be smaller. Yeah, it wouldn't be, you wouldn't be able to grow this fast, right? And so if you're, you know, part of a 5,000 person organization, you don't necessarily feel that same pressure that you feel, right? And so I think the organizations need to figure out how do you actually, you know, how do you create that pressure for that middle line, you know, worker? So they feel that.

And we need to do our part in that in making products a computer super easy to use. That's why we're launching workflows because the, you know, the example you had of you know how to prompt AI to do the fact checking on your articles, right? and you probably have a certain, you know, process that you use there that you repeat. For a lot of folks, they look at the open prompt, and it's terrifying. Yeah. They don't know. Like a blank page for a writer. Yeah, it's a blank page.

Exactly. It's the new writer's block. Right? The scariest thing you could ever look at. Yeah. And it's like, and you hear about, you know, I mean, all your reporting is like, oh, my God, AI is changing everything. I need to, you know, you need to be ahead.

You're going to get disrupted. And, you know, that's a guess. why we need something like workflows, which takes all these complicated scenarios and use case of AI and just breaks it down into a simple UI where you don't need to provide open-ended instructions, right? Right, right, right. Objectives. And so, yeah, so summing it up, the reason we're going to continue thriving in a very

competitive space is we're the best orchestrator and aggregator of all the intelligence, were the only AI company fundamentally committed to accuracy as like a core principle, and that's where we've made our big technology investments along with orchestration. And usability, which is really a design problem as much an engineering problem, it matters. And it's something that we've always had an edge in and we're going to keep innovating on. Yeah, well, the question is if these AI providers allow you to continue to use the models because they have shut down competing companies. So I want to take a break, and I want to go over that with you,

and then talk a little bit about the variety of models you do orchestrate, including the Chinese models. You have Kimmy K2 in there. So let's do that right after this. Look, if you have a kid in school right now, you know the drill. What should take 20 minutes of homework ends up taking two hours and usually ends in tears. And every good tutor, well, they're fully booked for months.

This episode is brought to you by Brainley. Brainley is an AI-powered personal tutor built by educators, not a general purpose chatbot. It doesn't just give your kid the answer. It walks them through step-by-step explanations so they actually understand the material. It learns how your child learns, diagnoses when they're struggling,

and builds a personalized learning path in under three minutes. Available 24-7, there's no scheduling headaches, and it's just a fraction of the cost of a private tutor. Finals are coming. Build your teen's study plan now. It only takes minutes. Go to brainy.com, slash big tech to get 50% off your first brainly subscription with my code big tech.

That's B-R-A-I-N-L-Y dot com slash big tech. Most leaders know how work is supposed to happen, but when it comes to how it actually gets done day-to-day across tools, teams, and handoffs, they're mostly guessing. That's exactly the problem Scribe Optimize was built to solve. Trusted by over 80,000 enterprises, including nearly half of the Fortune 500, it gives leaders a live into how work is really happening across approved business apps without interviews, manual process mapping, or extra effort from the team. And because it's continuously analyzing real workflow activity, the insights stay current instead of going stale the moment a process changes. You can see

which workflows are happening, where time is going, and which tools are involved. It automatically surfaces top issues, explains why they're happening, and even recommends ways to fix them with estimated time savings. And importantly, it's built with privacy in mind. So activities only captured in admin-approved business apps and user level data is anonymized by default. The kind of visibility that used to take months, now it's just always on. If you're ready to stop guessing and start seeing, visit scribe. how slash big tech. That's scrib-e.com. Insurance isn't one size fits all, and shopping for it shouldn't feel like squeezing into something that just doesn't fit. That's why drivers have enjoyed progressives name your price tool for years. With the name your price tool, you tell them what you want to

pay, and they show you options that fit your budget. Enough hunting for discounts, trying to calculate rates, and tinkering with coverages. Maybe you're picking out your very first policy. Or maybe you're just looking for something that works better for you and your family. Either way, they make it simple to see your options. No guesswork, no surprises. Ready to see how easy and fun shopping for car insurance can be? Visit progressive.com and give the name your price tool a try.

Take the stress out of shopping and find coverage that. fits your life on your terms. Progressive Casualty Insurance Company and Affiliates. Price and Coverage Match Limited by State Law. And we're back here on Big Technology Podcast with Dmitri Shevelenko. He's the chief business officer of perplexity. Dimitri, this is a really great, rich conversation. I appreciate it.

I've written about this. One of the big problems with all these AI use cases converging is that it used to be for the And for these big AI model providers, they'll build the, they have the demo products like the chat chit. This is the previous way of operating. And they'll offer their model that you can, you know, pay for intelligence and build whatever you want on top of it. But as we get to this style of Vagentic use case where everybody wants to build this stuff,

some will not be competing. But there's interest to have their own products like Claudeco work, like Codex, be the sort of system or agent of record, so to speak, that hands. handles all this stuff. And I think they might even prefer a world where, you know, that would just be the single app to rule them all. You're orchestrating their models. So long term, aren't you sort of at least dependent on their benevolence to allow you to use these models, even as you compete with their core products now? Yeah. I think ultimately all these companies are platform businesses in addition to product businesses. And they, you know, they aggressively petition us to use

their models. They want, they give us early access. They want us to run e-vows. And so we have, you know, the exact opposite dynamic right now where they're, you know, more than happy to take revenue from us. And, you know, they're the beneficiary of, you know, more consumption of computer credits as well. And I think they, you know, because they are all competing with each other on their platform businesses as well, and, you know, there's open source, which is, you know, continuing to push at the frontier, not necessarily at the frontier, but pushing at it. All those competitive dynamics are very healthy for us. Now, I agree with you, if we lived in a world where there was just one frontier model that was twice as good as the next best model, that would be a bad

scenario for perplexity. I wouldn't deny that. But since this industry has kicked off, there's never been a moment where the delta between the best model and the second best model was like more than maybe like a 10, 15% gap. And again, like, best model is probably I shouldn't even be using that phrase because it's best model at what, right? Yeah, there's, you know, it's the sub-specialization, right? And so the specialization is also a hedge against, you know, those sort of competitive dynamics.

And so I don't, I lose more sleep about, us preserving our execution velocity and continue to build our culture and our company through the intensity of this space rather than, you know, us getting cut off scenario because I'm not seeing indicators of that. If the models, your example of the model's sort of competitiveness is very interesting. I mean, we're at this point where the models are very smart, right? We have Anthropic, for instance, won't release Mythos because it believes it's too intense for cybersecurity. Great marketing, by the way.

Do you think it's marketing? No, I'm saying regardless of whether it is or isn't, it is great marketing. Do you think it's mostly marketing or truth about the product? I think. I ask everybody this, so I'm curious. I don't, I think everyone will have their own. I don't think we don't have access to mythos. So I can't speak to it out of, you know.

No, first-hand exposure. Yeah. But the people you speak within the industry, believers or mostly? I think there is a, I think what is a real concern is that models will be better at exploiting cyber vulnerabilities than they are at fixing them, right? Just like you can find these problems in the consultant presentations. Yeah. So I think that arbitrage, I think that's a real concern. I think that has already, you know, but I don't know if there's been some new capability that, like, didn't already exist.

I mean, you've been noticing, like, there's been more hacks and things over the last few years, you know, before mythos. So, like, I think this has been building up for a while. I guess, like, the, there was a long wind-up on my question to say, isn't there going to come a point where these models are just all kind of smart enough and compute becomes a commodity that, like, right now we're in this build-up? And eventually we just see parity among models, even though they're unbelievably smart and just like a lot of compute infrastructure and then sort of a price war that brings the price of all this stuff way down. Well, if a – It'd be good for you. Yeah, that'd be good. I mean, that's like in that scenario because, again, open source would catch up too, right?

But again, like you start – if we reach some kind of plateau, then you'll actually see even – you know, the local inference becomes more relevant because there'll be more investment there. I think it's really hard to make long-term predictions in this space. I'm fond of saying that the thing I'm most confident in is that six months from now, I'm going to personally have a perplexity, a top three priority, that today I don't know what it is. And the model companies themselves, you know, when they're baking the cake of a new model, like they don't know what it's going to taste like until it comes out, right? Meaning the capabilities, like when you train a model, you're not necessarily training it.

You know, you're making improvements, but you don't know exactly what the new capabilities are until it's out there and people start using it. And that is, you know, in some ways it's, you know, that's a core skill we've developed to perplexity is like zeroing in on when a new model becomes available. where is the, you know, actionable value for a user? Yeah. I mentioned this before the break, but you use the Chinese models. Kimi K2 is in perplexity.

I don't see DeepSeek in there anymore. So to clarify, we never integrate into perplexity any product that, or API, that is hosted in China. We have ourselves post-trained. We have post-trained. And open source models that come, that they're developed by Chinese labs. We run those in U.S. data centers. We post-trained them for accuracy and removing, you know, things that are not accurate from them. Well, like, you know, different countries might have, you know, certain political agendas that they try to integrate into models.

And you find those in the models? I mean, we've published some research on that with Deep Seek. if you go back to it. What to answer questions on Tiananmen Square? Yeah, there's those sorts of now again, like that's, you know, we also solve for that with grounding with accurate search, right?

And that ends up, you know, if you're using the model fundamentally for reasoning, that becomes less of an issue. But it's really impressive what the Chinese labs are doing and the progress they're able to make. I think open source is good overall for users.

is ensuring that, you know, pricing remains competitive. And obviously there's more we can do in the post-training space on an open model than a closed model. And so that lets us kind of, you know, accelerate our work around accuracy, conciseness, you know, adhering to certain task workflows. When Jensen says it's important for the entire world to have their AI built on a Western or U.S. AI infrastructure stack, if you could do what you just did, what you just told me with Kimmy K2, which is down the weights, post-training it the way that you want, why does it matter where the models are developed? What does it matter if, let's say China has the lead in open source? what would be a bad scenario is say that the best open source models, their architecture is done in such a way where they don't run on NVIDIA chips. They only run on Huawei chips, right? So that the kind of, I think the scenario Jensen is concerned about, rightfully so, is where, you know, software drives the hardware cycle, right? and where, you know, imagine the flip of the scenario where right now Chinese companies are trying to get access to Nvidia chips because that's where the model architecture is, right?

And they need the Nvidia chips to be able to run them in an efficient way. What if it's flip the other way around where, you know, it's the Huawei chips are the ones that U.S. companies would need to get, right? Oh, that makes a lot of sense. So then China can export control the U.S. and control AI. Yeah. So I think that's the, I think that is, you know, when you have this like... Why didn't he just say that in that Dorcasch interview? It's like a very straightforward answer anyway.

The, well, no, so Johnson is very good at Com, so I wouldn't, you know, I think there's a new, I mean, there's certain things he can't say probably too that, you know, can't say certain names. Yeah, we can say it here in the show. That's fast, but the model, Chinese models are good. they are you know they're pushing the frontier they're not at the frontier but they're pushing it yeah i want to end here there is this interesting argument i think you're you have a perspective on it at perplexity that this is a great article from cnbc that dea trobeza wrote AI demand is inflated and only anthropic is being realistic i think that the crux of the argument is that like people have been running massive amounts of work of of of workflows on these like

$20 or $200 a month plans and you know they are there's like a lack of ability to serve them and so therefore these AI companies are showing immense demand and going and raising money based off of it where like everything's going to change once you have to actually charge per token as opposed to unlimited. Like you wouldn't do an unlimited electricity plan or an unlimited fuel plan, but for some reason, a lot of these companies have been doing this. Do you think that this is like a legitimate issue that she's pointing out that basically like we don't really know what AI demand is because it's been subsidized so heavily for so long? And if so, what's the answer here? So we, at perplexity, we've never subsidized paying users. So if you're on a, you know,

or Max Plan, you know, thank you. You're contributing to our success. You're welcome. And, you know, we see great retention, so clearly folks are finding value there. And that's actually why computer credits are so important, right? So that as you have, because you can have a certain computer tasks cost you $50 for, you know, say it's like video generation and it's like long horizon running. you know, one task can cost up to that much. And then you have certain tasks that costs, you know, five cents.

And so there's no way to encapsulate all of that in a, you know, subscription product, right? So I think the mental model I would have is AI is going to become a lot like Costco, where you pay for the membership, right? And that gets you in the store. And that's actually the part of Costco's business that is, you know, the highest margin. and then you have, you know, everything you're buying in the Costco, you know, you have confidence that there's like a max margin, right? And those are kind of like computer credits, right? And it's, you know, some people go to Costco and they just buy the hot dog.

And then, you know, there's people who go and spend, you know, thousands of dollars every trip. And that depends on their needs. You know, but I don't, I think she's reacting to some to, I think it was cursor kind of advanced this, this date. a point that, like, Claude Code was subsidizing, you know, a subscription tier. I think that will normalize over time. But the behavior we're seeing with computer credits where, like, people are paying for usage, right? Like, there's no, there's no subsidization.

There's no kind of breakage that's driving it. And finding value and paying more every month. As they use it more, I think we're. Yeah, I think it's a safe investment in all the compute and data centers. Okay, really the final question. I mean, how do you keep up? Like, perplexity has been, I would say, early on three trends, right? AI search, AI browsers, and now this computer use must be tough to set strategy as a company

with things changing as quickly as they do. So what is the process that perplexity uses to make decisions about, you know, strategic direction and product plans, you know, with all these capabilities just like kind of blasting all the time. Yeah. I think part of it is keeping a very lean team. You know, as we've increased our ARR by 5X, you know, from $100 million to $500 million, we only grew headcount 34%. You only have 300 people.

Yeah. That's crazy. So, you know, that is, and I mean, this is what I try to share with companies outside our walls, is, you know, you're going to be, you know, the world will keep changing faster. And so your only way to adapt to that is to be quick at making decisions and not like, you know, tying yourself to one path. That's also a lot of the, you know, not to bring it back to why perplexity computer is great. But you don't want to be, you know, tied into one model if another model is going to be better three weeks from now, right? the world is very unpredictable.

And so you want to have agility and you want to make quick decisions and be willing to revisit your decisions, right? And, you know, I think, you know, I think having the humility of not knowing what the world's going to look like two years from now is a big part of being successful in that world. Yeah, I mean, I wrote a book with this title, but it is always day one. Really, really sort of felt that way beforehand, but in this world, you can't be tied to any legacy. You have to just basically see what the new is today and how it works and take charge. And you guys have been good at doing that. Thank you. It's great to see you again and thank you again for coming on the show.

Hopefully we can do this again soon. My pleasure. Thank you. All right, folks. Definitely check out the link in the show notes for the 618 event. We'd love to see you there. And until then, we'll see you next time on Big Technology Podcast. Instacart knows that some people go bananas about getting the perfect, well, banana.

Some want them green, some want them ripe, some want them ready right when they hit their doorstep. But with Instacart's preference picker, available at most retailers, you can choose to get your groceries just the way you like. That means perfectly ripe bananas, deli meat sliced just the way you want, and avocados that aren't still hard as a puck in the third period. So don't cross your fingers and hope for the best. Download the app and get groceries just how you like with Instacotons. chart.