Transcript 0:01 [upbeat music] Hello, everyone. 0:19 Welcome to The Digital Leader Show, where we talk key issues and news at the intersection of business and technology in hopes of making you a better digital leader. 0:28 We're live every Wednesday at noon Eastern, 9:00 Pacific on LinkedIn and YouTube, and the audio recording is available wherever you get your podcasts. I'm Dan Goodstein. I head up ERPA AI. 0:36 We're a professional association, media and peer-to-peer research network providing education, networking and advice to over 130,000 members globally all around automation, enterprise AI and digital transformation. 0:48 Today we're discussing AI product building for enterprises and solopreneurs with Mozilla's VP of Product, Raj Singh, plus headlines for digital leaders from the world of enterprise technology. 0:58 Thanks to our sponsor, Oracle NetSuite. Their cloud ERP helps provide real-time visibility into operational and financial performance. Podcast followers can get preferred pricing, up to 45% off at erpaai.com/netsuite. 1:11 My co-host today is Chris Sirdak, senior analyst for ERPA AI, award-winning author and former White House Chief Transformation Office advisor, partner at Gartner and practice lead at EY. Chris, good afternoon. 1:24 How are you, sir? Into September. Can you believe it? Welcome to September. It's, uh, feeling like fall over here in New York. The kids are back to school, the wife is happy, and, uh, back to work. Back to work. 1:37 I'm jeal- I'm jealous. September in, uh, in Southern California is monsoons, humidity, and heat. It's a, it's a hot one out here now. 1:45 Well, you know, we, uh, often complain about the weather in New York, so it's only fair that you get to complain once every few years too. 1:51 Yeah, I wor- being a New Yorker, I know the way you know when New Yorkers are always complaining about the current season and wishing for the next season. [laughs] Right. No matter what. Yep. I've always... 2:02 My colleague's always like, "Why don't you just move to Florida already and stop complaining?" [laughs] Right. Exactly. One season. [laughs] So, uh, just a la- Labor Day news. Labor Day news. 2:11 Kraft and Heinz are breaking up. Hot dogs are breaking up with American cheese. Breaking up is hard to do. I mean, you know, now you gotta get your mustard and your ketchup from two different places. 2:20 What are we gonna do? 2:22 Uh, it's interesting to me, you know, I, I'm sure there's a, a, a smarter person out there who can explain this to me, but I, I just feel like, you know, all we do is cover big acquisitions, big companies, you know, eating up, uh, smaller companies, and then at some point, five years later, 10 years later, just rollout, reorganization, divestitures. 2:44 It's just like, uh- Sure... no matter what, it happens every week. Yeah. Berkshire Hathaway, um, what's his name again? Whatever his name is. You know, he- Buffett... 2:52 they're the number one hold- yeah, Buffett, number one holder in, in their shares, and he was the one that was super bullish about the merger and so forth, and I, I saw some quote from him saying, you know, he's really disappointed it turned out this way. 3:04 I... You know, the, the, the, that era of just constant conglomeration and bigger and bigger and bigger and efficiencies, it's run its course. I don't know if you recall, but in one of my books, I call it Snickering. 3:15 You know how, like, the Snickers bar at Halloween, they keep getting smaller and smaller and... but, but they keep the price the same. Yeah. 3:20 And eventually it's so small you can't put a peanut in it, and now you have a m- a Milky Way. I, I think- Sizeflation? Is that what they call it? Sizeflation? Yeah, or shrink- shrinkification or something like that. 3:30 Yeah. Yeah, yeah. Shrinkage, right? So I think Kraft Heinz just Snickered themselves out of existence. And, and guess what? We're not, we're not gonna cost cut our way to greatness anymore. 3:37 We're gonna actually have to- Mm... actually have to go to work and do some work. Yeah, yeah. 3:42 Um, meanwhile, uh, uh, your, your buddies at Anthropic raised, uh, another 13 billion, with a B, in case you're not following this. So- $1 billion... 3:51 so they're at a $183 billion, uh, valuation on, uh, I don't know, a couple of hundred million dollars in revenue or something like that. 4:00 [laughs] They, they need more fricking sharks with fricking lasers on their fricking heads, I guess. I mean, I... God only... You know, more scale. It doesn't work, but maybe, you know, let's do it more emphatically. 4:09 I, I'm, I'm just hopeful that at some point someone realizes that LLMs are lovely, they do amazing things for very targeted reasons, and maybe $5 trillion isn't a great idea. I don't know. We'll see. Yeah. So sorry. 4:23 So, so Claude, uh, Claude Code is, is, is, uh, is, is 500 million in, in revenue, but the, but the company's, uh, run rate is up to, uh, just over, uh, u- u- almost five billion actually, up from, up from a billion. 4:36 So you know, 300,000 businesses on the platform, uh, but still a, a sevenfold increase on accounts generating over 100K they say. Yeah. A billion here, a billion there, eventually you got a recession, so let's go. 4:51 [laughs] I... More power to them. I hope it actually works. I mean, we, we, we look, we see the results. This isn't a surprise to many of us who have been in tech for a very long time. 5:01 It's not different than every other wave, like we keep telling ourselves. Um, and, and, and I, and I hope we can somehow come up with a soft landing. I, I know I often come off as a naysayer and so forth. 5:11 I'm just kind of a realist. It's good for certain things. Is it, is it $4 trillion good? Eh, hard to see it. Uh, US government is, uh, doubling down on AI. 5:24 Microsoft announced a, a GSA major AI for government contract, uh, this week. I don't know what the, uh, revenue was, but it's, it's gonna get Microsoft Copilot, um, and, uh, Azure and a, a bunch of... 5:39 Dynamics, a bunch of other, uh, government, I guess a government version of, uh, Office 365- And can-... to, uh, government employees... 5:47 a- and so I say that, Dan, and then you immediately ma- make me, immediately make me put my foot in my mouth, right? Um- I try. I, it's hard. I try- Yeah... every week and I usually fail, Chris, but... 5:56 I mean, the thought of, of Copilot helping the IRS review my taxes is absolutely horrifying. Um, but apparently that's where we are. So you know, keep, keep good records, people. 6:07 [laughs] 'Cause I, I'm not, I'm not thrilled about the prospect. But by the same token, we need to be testing the technology. We need to be understanding its limits and its applications. 6:16 So- My only, I guess my greatest concern is that we've taken the, the Petri dish of experimentation with AI and we turned the, our entire society into that Petri dish. So let's hope the experiment goes well. 6:27 [lip smack] Uh, the other big headline before we bring our, our guest on, uh, to today's show is, uh, got, got some, uh, feedback on this, uh, Google antitrust, uh, hearing, uh, or case. 6:42 Uh, and it looks like the, you know, the, the shakeup that maybe some of us were expecting may not be as, uh, uh, as shaky. So, uh, basically the judge says they don't have to sell, they don't have to divest with Chrome. 6:55 So, uh, who's that, who's that company we covered a couple of weeks ago that made that, that wild, uh, offer on, on, on Google Chro- So that's not gonna happen, but, um, uh, they're gonna have to play nicer with, uh, in, in the ad space. 7:07 So, so a lot, a lot of people are saying not, not the big change that, uh, some were expecting or, or hoping. 7:14 Um, it, it, it's reminiscent back to what we talked about, I think, a couple of weeks ago, like the, the Microsoft thing, like a slap on the wrist. 7:20 Let's, let's, uh, let's play nice everybody, but no, no major revolution, uh, in the tech industry here. 7:28 Well, gi- given I have a bit of a legal background, I suspect what happened is that, um, the judge went back into chambers and then googled whether or not Google is a monopoly, and Google said no. 7:37 So, so, you know, hence the result you get. [laughs] Did they Google it or did they, uh, ChatGPT it? They Googled it. Yeah. 7:43 If they'd asked ChatGPT, the judge might have gotten a slightly different answer, but it's a little late for that, apparently. Funny. Funny. 7:49 [laughs] Well, let's get to our main topic today, which is, uh, developing AI products. We have today with us Raj Singh. Raj is currently the VP of product at Mozilla. 8:00 He's a seasoned entrepreneur and product leader with a background in AI-driven startups. Raj, thanks for being here. Thanks for the, uh, opportunity. 8:10 So, uh, before we jump in, give everybody a little bit about, uh, your journey, because although you're at Mozilla now, you've, like as I mentioned, you've been in the startup world. 8:17 I think you've sold a couple of companies. So you've been there, done that. Give everybody a little of your, your, your background. I've been, um, doing startups my whole career. 8:25 You could probably argue I've been doing startups before it was popular, [laughs] um, and maybe not popular. I don't know where we're at today, but, uh, um, uh, yeah. I grew up in the Bay Area. Uh, California native, 8:38 um, uh, or Bay Area native. Uh, dropped out of, or sort of in masters, started my first company, uh, doing file sharing back in the P2P era, if you remember that. And then, uh, and then really kind of never looked back. 8:49 Uh, had a dating site, had a for ca- a college dating site, had a very popular ringtone application. Um, later did live video. 8:58 Uh, almost, I'd say thematically, um, everything's been consumer-facing, so I tend to focus on personal pain points. 9:04 I find that much easier to sort of identify with the problem and, uh, gives me that e- extreme, like, sort of domain experience and understanding the empathy. Um, had, had some venture-backed companies. 9:13 We've had four exits. Uh, most recently, uh, was doing a meeting summarization. I had a startup doing meeting summarization called Pulse, uh, that was acquired by Mozilla, and that's where I'm at now. 9:22 Uh, and I'm responsible for, uh, new products focused on, uh, small business. So Raj, a serial masochist, startup guy. You know, uh- [laughs]... 9:33 you know, there, there's a lot of jokes, uh, around startups, uh, and I pro- I probably could spend 15 minutes running through them. 9:39 Everything from like, uh, uh, you know, um, if you know a lot of VCs, uh, you must suck at fundraising because you're pitching a lot- Depends on the amount. [laughs]... 9:47 to, uh, you know, uh, have a kid while you're doing a startup because you're not sleeping anyways. You know, [laughs] there's a lot of jokes like that, right? 9:53 Uh, uh, but yeah, it's, uh, it's certainly not the best path to get rich if that's your sort of end game. 9:59 Uh, but if you really, um, if you have a passion on the topic, like, I mean, I think this is, like, probably the, one of the most common questions, like when's a good time? It's never a good time. 10:08 It's, the time is now, right? So you're gonna regret- It's better, it's better than a life replacing TPS cover sheets. I have no regrets. 10:15 [laughs] I've, I've, uh, I've built a, a range of products, and I've always built them to solve my own pain points, and I always bank that if I can solve it for myself, there'll be other people who can benefit from it too. 10:25 Awesome. So Raj, help me, help me understand your role, you know, to the, to the outsider onlooker. I think, okay, Mozilla, Firefox, like what, what else is there? 10:34 But you're, you're responsible actually for, for new products there. What does that mean? Yeah. So, uh, most people obviously think of Mozilla, they think of Firefox. 10:42 We've done a lot more than just Firefox over the years. Uh, you know, helped create Rust and a range of, a range of amazing things. Um, but 10:50 we also have, uh, new products, uh, that we've been building and working on, and part of that is search is changing, the browser landscape is changing. Um, so we need to diversify, right? 11:01 We don't wanna be completely dependent on... I heard you earlier talking about Google, but, like, Google's a large chunk of our revenue. That's obviously all public information. 11:09 Uh, we don't wanna be completely dependent on, on sort of one, uh, one source of income. So we're diversifying. As part of that, uh, we felt that a great space to look at was small business. And why? 11:21 Well, the fastest growing, uh, uh, business segment in the US is the freelancer. It's the solopreneur. And I think part of this is sort of the post-COVID effect, uh, what I like to call the ownership economy. 11:31 People wanna be their own boss. Uh, people are enjoying, uh, controlling their hours, uh, integrated lifestyle between, uh, family and work and whatnot. 11:38 And so, uh, and so then we then spent energy, you know, we've spent energy kind of really digging into what that means and then identifying like, hey, what sort of tools that we could build that are sort of gen AI first, or agentic, or however you wanna describe it, that can help make life easier for those solopreneurs. 11:55 So, um, that's what I'm focused on. 11:57 Um, there's other initiatives too within, uh, Mozilla around new products, but a lot of this again is under the macro umbrella of trying to diversify our revenue, uh, beyond just sort of just simply search in Firefox. 12:10 So that's interesting. 12:11 Like, you know, as you, as you think about products, you know, I, I feel like a lot of what we talk about is enterprise, but, um, you know, how do you think as a product guy, how, how do you think differently about products that are designed more for small business or the solopreneurs, the gig economy versus, you know, big enterprises? 12:28 Like how, how does that, how is it different when you, when you start ideating ar- around it? So I, I, I think it's vastly different, right? 12:35 So I always tell people- Um, I have a lot of respect for people who do B2B because I think it's very hard to understand and empathize when you're not the primary user. 12:48 Uh, you know, you have to go to a buyer, and the buyer's telling you, "This is what I want," and blah blah blah and whatnot, and it might be in a vertical or a domain that you don't know a lot about. 12:56 I think it's really hard. Like, I think that takes a special set of skills, uh, to do that. Um, when you are the primary user, you really understand the pain points. You're like, "Okay, this is what I need. 13:06 How can I fix and solve it?" Right? And so I think, you know, and I think that really highlights a core distinction between, uh, consumer products and enterprise products. 13:16 Consumer products, we're building for the user, and enterprise products, you're often building for the buyer. And there's a lot of examples of that. I don't wanna go into names. 13:23 If you're focused on enterprise, you know exactly what I'm talking about. 13:25 But you see many instances of, uh, applications, uh, some of which, uh, that we use even internally, where it's not particularly the best user experience or it's, it's, it's painful. 13:35 And I, I, I sometimes joke that the number of pages in your internal wiki, uh, uh, graft versus the tool is sort of a indication of how user-friendly it is. 13:44 If you have, like, tons of support pages for that particular tool, it's probably not very user-friendly, right? [laughs] Most of the time. All right, so- [laughs] But, but, you know, when you're a consumer application, 13:54 it's not forgiving, right? You, you, uh, you install this thing and you get confused on step two of onboarding or, like, at the homepage or whatever, you just bounce. You drop off, right? 14:02 And you, you lost that user, and you may not be able to re-engage that user for 18 months or whatnot when they hear about it again or whatever. So it's a very different way of thinking. 14:10 Now, of course, with consumer, you know, consumers are often less willing to pay. It's a higher bar. It's much more user experience focused, but it's, uh, you know, it's a phenomenal, um, beachhold once you have it. 14:23 Uh, you know, once you have a large consumer audience, and then you can pad on an enterprise business, it's amazing. Uh, because now you're extremely diversified. 14:32 Uh, you can sort of, uh, you know, you're defensible against all sorts of, like, macroeconomic sort of things because, you know, a consumer base is representing hundreds of thousands of people subscribing, right? 14:41 Look at Calendly. I think it's, like, 60, 70%, uh, literally consumer subscribers, uh, and then the remaining, uh, you know, as business subscribers, uh, you know, amongst their sort of whatever, two, 300 million ARR. 14:54 They're totally defensible. I mean, in this world where scheduling has become commoditized, you would imagine they would just get crushed. Right. 14:59 But they have so much consumer traction, it's much more difficult than it may sound, and we see this all the time, for example, like social networks. It's virtually impossible to disrupt. 15:09 [laughs] If, if, and man, this is actually fascinating because if you think back to, like, the 2000 timeframe as we were migrating from Web 1 to Web 15:16 2, you had, you had players like, um, BA- BAE Systems, the enterprise integration layer that would tie... Because everyone, Web 1, everyone was an island. And so to jump to Web 2, we had to tie them together. 15:27 Now it sounds like what you're trying to do is everybody's a solopreneur, love the term. 15:32 As we jump from Web 2 to 3, those people have to have a way to tie back to the mothership of their, of the companies they're serving, like the, the parents that's paying them and so forth, and, and have that business integration. 15:41 Is that sort of what you're attacking then? Yeah. I'm, so I'm, I'm, I would say even simpler than that. And just to be clear, like, my interpret- my version of Web3 is actually a slightly different definition. 15:51 My version of Web3 is much closer to Web 1. 15:53 Like, I think we went from closed, like AOL walled garden, to open, where everybody was connecting and integrating with each other, to basically APIs now that are all pay for play, require partnerships, login registration for all content, content walls everywhere, right? 16:06 And, uh, and I get that. That's because of LLM indexing the web- Mm... um, people becoming more protective. You know, LLMs are basically aggregators. They're not surfacing the sort of authentic raw content itself, right? 16:16 And so, uh, the whole ad business is being, been thrown upside down. Uh, no, we're starting super simple. 16:22 You're, you're, um, you're a coach, let's say, like a basketball coach, and you wanna s- you wanna grow your business. 16:27 Now, usually a basketball co- or like any, any sort of solopreneur, they don't sort of, like, uh, hit that point, like, overnight. It's more like they were doing something on the side. 16:36 Maybe they were helping coach or something, or maybe they had an Instagram page and they're posting some videos of some private coaching they're doing and they're... And somewhere at some point they go, "You know what? 16:44 I wanna scale this thing. Like, I wanna, I wanna make this bigger. Like, what do I need to do?" Right? Like, my Instagram page isn't good enough, or my Facebook page or whatever isn't good enough, right? 16:52 At that point, maybe they need a website, right? And so they go and sort of set up a website, and they're like, "Oh, I wanna make it look more official. Maybe I want a business email." I mean, I don't know, right? 16:59 Then they're like, "Oh, you know what? I need to do more social media. Like, I wanna be able to, uh, you know, share more clips." So maybe they go and set up a TikTok account. 17:07 Maybe they set up a Twitter account, maybe a Threads account, whatever. Who knows, right? Right. So now they need maybe some tool to help them with that. Maybe they launch a blog and they want a newsletter, right? 17:15 Then maybe they go set up a Substack. I mean, I don't know, right? 17:17 They want, like they have a list of people and they wanna be able to re-engage them, or it could be, like, something for SMS, you know, so they can text message all their clients and say, "Hey, we're running a new summer camp this summer. 17:26 Do any of your kids wanna play basketball?" Whatever, right? Um, now all of a sudden they've got income and they've got expenses at a bigger scale, so now they need some accounting tool, right? 17:34 Like yes, they maybe they're going and buying QuickBooks Self-Employed, but it's, you know, these complicated applications. These are not simple apps. These are not very user-friendly. 17:40 They were kind of designed for larger accounting firms, right? You need, you need integration and orchestration of all that stuff, otherwise you're gonna be buried. Yeah. 17:46 And then now all of a sudden, you know, uh, maybe they're, they're busy, uh, uh, setting up all this coaching and they have a busy calendar and they need a scheduling tool. I mean, I don't know, right? 17:55 So we're kind of going through, like, ranges of use cases. There's a whole sort of cluster of different workflows here. And so then the opportunity then goes, like, okay, let's look at this today. 18:04 Let's look at what exists today and what their workflows are. And if you look at most, uh, small business tools, they're targeting small business. They're not targeting solopreneurs. 18:11 They're targeting businesses that are at least 10 or 15 or 20 people, right? And so they're oriented in that fashion and, and so there's, like, kind of upfront learning time and whatnot, right? 18:19 But when you're targeting a solopreneur, it needs to be dead simple. The UX needs to be super easy, whatnot. And their pa- their pain points are a little bit different, right? 18:27 Um, so, so then we look at all this, then we go, okay, well, how could we make these experiences awesome, simple, and, and just more amazing because of GenAI? 18:35 And so GenAI is kind of the disruptive unlock here, uh, because we can literally make things almost programmatic or almost agentic in some cases. 18:41 So one of the first products that we launched within Mozilla is called Solo. Uh, it's an AI website builder for service providers. 18:47 And so you literally can say what your business is about, you can point to your existing social media, and it kind of pulled together your initial website. 18:53 And then you can then, of course, further edit it down and whatnot. We have new tools we're launching in the spaces of social media, uh, blogging, uh, virtual... You know, like a smart voicemail. 19:03 You know, if you're that plumber and you're working and you can't take that call, and you need, like, something to capture your lead. Uh, back office accounting, uh, and more, right? 19:10 And so we're looking at a whole suite of tools here to enable the solopreneur, and we're really excited about this segment because we think it's just generally been an underserved segment, and it also plays really well with sort of Mozilla's mission around sort of democratizing the internet and sort of helping, helping the underdog. 19:24 And is that international? Ju- I, I would assume so. Yes. Yeah. I mean, tremendous opportunity in the developing world for that. Absolutely, yeah. And I mean, we're... 19:32 You know, depending on the product, um, in some cases we're launching globally on day one, in some cases we're maybe, you know, launching in specific segments because it's more difficult to scale, but the intent is to be available worldwide. 19:46 So we, we, we talked a little bit before you, before we started live, Raj, about, you know, just the, the AI thing, right? Mm-hmm. 19:54 How important is it for a, for a solopreneur to, you know, have gen AI skills, prompting, that kind of stuff? Or is that gonna just be something you hire somebody to do? So I think, I... 20:09 To me, that, that's the wrong framing. So I think, I think if you're hiring somebody to do that, then we're already at, in a bad spot. 20:15 Like, [laughs] like, so the way I think of gen AI, uh, and the way I like to describe it is it's democratizing fractional expertise. 20:23 And so we humans have had all this creative energy that's been bundled up, but we can't necessarily express that energy, and the reason is we couldn't do it. 20:30 Like, you know, maybe I have this, like, incredible idea of a, of a, a home I wanna design, or maybe I have this incredible idea of this new, like, object or whatever- [coughs]... 20:38 this toy or this storybook, uh, maybe this illustration, maybe this piece of software. I don't know. 20:43 Previously, if I wanted to pursue any of those things, I would have to either spend enormous amounts of time to learn various tools and, and, and do things that are not necessarily core to my craft, um, or I'd have to go find specific experts to help me with those things. 20:56 But now with ChatGPT or Gemini or whatnot, you know, Claude, whatever you're using, we have things that can help you get there, and that's amazing. I mean, this is a... 21:05 This has been one of the greatest sort of transformational unlocks, in my opinion, um, uh, in, sort of in my lifetime. I mean, probably mobile was one of them, internet was another one, right? 21:15 Um, and all of a sudden, we have all this creative energy that's getting unleashed. People are building things again, right? 21:21 And every product that has any form of creation, we're seeing this amazing trend where vibe is becoming this new modality or new interface into these products, right? 21:30 So I- if I'm doing anything that involves creation, whether it's a- an application, whether it's a, a, a visual design, whether it's a, you know, a toy or whatnot, people are expecting to be able to chat into it. 21:40 People are expecting to be able to, like, further iterate and describe where they wanna go. And, and I think all of that's all sort of amazing. So now we take the solopreneur. 21:48 You know, historically, they were very dependent. They had to go and ask people around like, "I need to go hire an accounting firm. I need to go..." Like, how do you use Wix? 21:56 I mean, Wix, as simple as it is, is a very complicated product 'cause these things are being designed for all sorts of use cases. 22:02 They're not just being designed for, you know, let me set up your simple business or, you know, maybe they're like, "Oh, I need to figure out, um, you know, how to send text messages." I mean, who knows, right? 22:11 And so now that we have gen AI here, I think this is just an amazing unlock. And so sort of my opinion is everybody, 22:18 uh, should get out of the mindset of hiring people for this, but everybody really needs to spend that energy to think about how they can bring gen AI into their workflow. 22:25 'Cause I think it's a necessary upskill that's happening across the ecosystem, and we are seeing, you know, uh, uh, I guess to say bluntly, you know, I'm not afraid of AI taking over human jobs. I don't see that at all. 22:37 I mean, companies are g- becoming more aspirational. Humans are providing judgment and taste. Uh, you know, we're directing what needs to be done and whatnot. And companies are gonna be chasing 10 trillion, right? 22:47 We're just all thinking too small, right? And you're seeing more, more boundaries getting, uh, uh, people bumping into each other. 'Cause Salesforce may now do search, and Google might not, might now do CRM. 22:55 I mean, who knows, right? Um, but, uh, you know, uh, um, the... 23:00 What, what is risky is if you're, if you're the IC or, you know, person in that company and you're not using gen AI, the chance that you might get replaced by somebody who is, that's much more real. 23:09 And so I think it's really important that everybody start, like, sort of spending time thinking about, um, if you're not already adopting, how they can further use it in their day. 23:16 And a good way to start, I often recommend to people, is just document your day, right? 23:20 Write out what you're doing in every 10-minute segment, and then kind of post it in ChatGPT and say, "Hey, how could you have helped me at each of these steps?" 23:26 And you'll realize like, "Whoa, you could have helped me here, and you could have helped me there," or whatnot. And, uh, and that's just a great way to start. 23:32 I- if, if I may, Raj, it sounds like you're, like, not to say you didn't know this before, but it, it's a rediscovery of something we learned in the RPA days 10 years ago, which is task-based automation is amazing. 23:43 But the notion you're gonna, you're gonna automate away entire chunks of humans or entire processes, that, that's an un- unrealistic expectation. Is that kind of fair? 23:52 The task stuff is great because y- it's very definable. You can get the, you can get the value that you need. Sure. And then you orchestrate or maybe you get some help o- orchestration. 24:00 But this notion of I'm gonna replace entire humans and entire human functions is, um, is kind of, kind of the move. Yeah. So I, I think it's case by case, workflow by workflow, right? Yeah. But it's changing. 24:10 But my sort of point is the new hiring is because the companies are thinking bigger. Mm-hmm. It's not like, "Oh, I can get this done with less people. We're gonna slow down." No, I'm gonna still... 24:20 [laughs] "I wanna go faster now," right? "I wanna do 10X as much work." Um, the reason, uh, the reason humans are still in the loop or always will be in the loop is I w- I, I, I have a term for this. 24:29 I call it, uh, designing for 95% UX. Mm-hmm. And what I mean by that is AI historically... And I've been in the AI space for a long time. 24:37 Actually, two startups ago, I was the first.ai domain on the web with my calendar startup Tempo, which I, uh, spun out of Stanford Research Institute with early Siri, you know, uh, type stuff. Uh, anyways, um- 24:48 The reason is AI has consistently been 95%. So if you think about it that way, that it's 95% accurate, how do you design a user experience that's accommodating? 24:57 Uh, you know, a good way to think about that is if I was designing, um, like a website, 95%'s fine. Like if the content is 5% off, I can further edit it and make it perfect. 25:07 But if I was doing, let's say, we talked about scheduling earlier, if I was doing a tool to automatically schedule my meetings, if it got it wrong 5% of the time, that's disastrous, right? 25:15 Because that's a high, uh, opportunity cost, right? Mm-hmm. Like I'm not showing up to a particular meeting. I didn't make it to this podcast 'cause my AI got the, you know, event wrong in my calendar, whatever, right? 25:24 Now, of course, it's not, you know, it's maybe not 95, maybe it's 98, depending on use case or whatever it is. 25:29 But I like to use 95 because it forces you to think about, like, okay, how do I design a user experience to accommodate for that? 25:35 But, but Raj, the good news there is that the, you know, can- when your agent screws up, it gives you digital plausible deniability. [laughs] It gave me the wrong calendar invite, right? I mean... [laughs] Yeah. 25:44 We don't want to even, we don't want to even be there. So actually, interestingly- Yeah... 25:46 what you are seeing is, I would say when ChatGPT was initially released, for those that were not in this sort of AI category, many of us that were were seeing sort of the LLM stuff coming. 25:56 Uh, most of us were training, like with my previous starters- startup, we were training, we were fine-tuning the top just regular language models, not large language models. 26:04 Um, but, uh, what we saw was a surge of startups in 2022, 2023 sort of going out with a very simple pitch: "Uh, we're gonna replace this entire workflow with GenAI." 26:13 And then what's happened is they went out to market and they're, like getting stuck at this sort of 96, 97% threshold- Mm-hmm... or ceiling or whatnot. 26:19 They're trying fine-tuning, they're trying, you know, custom data, whatever, you know, who knows, right? 26:24 Uh, you know, further, you know, more context, whatever, and they're not really hitting that sort of inflection point. And so then what they're realizing is, hey, I need to bring a human back in the loop. 26:33 Um, and so that's changing the overall sort of ROI calculation of that startup. And so we're seeing a lot of startups adjust. And so as a result of that, some startups are real- are just not making it. 26:42 Uh, and that's in part because their original premise was flawed, because they couldn't get to sort of 100%. 26:47 And other startups have said, "Hey, we're gonna bring a human back in the loop to make this great," um, and the ROI is still great, right? 26:53 And so you're seeing kind of both camps, that divergence happen over the last year. I, I would often say the automation should supplement humans, not supplant them. 27:02 You know, I think, I think AI by very definition, uh, is, is doing incredible. I mean, you're, you're seeing, you're seeing more and more capability, right? 27:09 The fact that, like I saw a recent version of ChatGPT starting to say, "I don't know the answer to that," which in itself is quite amazing, right? Like that- Hallelujah. [laughs] That's amazing. 27:18 [laughs] That's, you know, they... I think there's something around kids and brain understanding or whatnot that they don't even, they can't even comprehend that kind of answer until some certain age, right? Mm-hmm. 27:26 That shows a certain level of brain development. 27:28 At the same time, you know, there is still judgment, there is still taste, there is still conviction, there's communication, there's all sorts of other things that are involved in, uh, driving something forward. 27:39 But, uh, uh, you know, right now it's a massive uplift. Uh, have we hit some kind of ceiling threshold? I don't know. 27:45 People are, uh, many people argue that ChatGPT-5 was not a significant improvement over previous versions. 27:52 However, if you look at benchmarks between the different competing, uh, you know, large language model, uh, the hyperscalers and whatnot, we're seeing, uh, still some benchmarks get blown away and, but in some instances it really has slowed to a crawl. 28:04 Uh, my gut feeling is, you know, we are gonna... We're, we've already hit the point where there's not enough differentiation between the different models. 28:12 Um, so whether you're using Gemini, whether you're using Claude, whatnot, like you don't really notice. At this point, it's good enough. Like it's hand-... 28:18 Unless you're speci- in a very specific domain, like you're doing medicine design, or you're doing, like you're specifically coding or something of that nature, like you're not gonna notice for your day-to-day activities much, much sort of difference. 28:29 So I feel like for the general consumer, we've already sort of hit that inflection point. 28:33 Um, I think, uh, more broadly, uh, historically, we haven't seen massive step functions unless we've seen massive step functions in compute. 28:42 So if you look at like, uh, you know, 2000 era, you couldn't really do machine learning, right? Machine learning was just way too compute intensive, right? 28:50 And so all of sort of the AI efforts pretty much through 2000 to 2010 were much more heuristic based. They were like, you know, semantic analysis, it was rules based, uh, entity recognition. 29:00 It was basically all basically some variations of rules and heuristics, right? 29:03 It wasn't really till 2010 era we had enough compute, you started seeing small, uh, machine learning based models, and then deep learning, and then language models and whatnot, right? 29:12 So we might be hitting another step function now that might ceiling, let's say, in the next some, some number of years, and then the next breakout will happen when we have the next breakout in compute power. 29:22 And so that could be either some kind of technology breakout in power or, or in like, you know, GPU, whatnot. I mean, I don't know. 29:28 Uh, and who knows how long that'll take, but that, you know, that, that's the sort of indeterminate. Jet, you jumped right over all the setting up of the Hadoop clusters we went through for like six, seven years there. 29:38 Oh. [laughs] That, no, I, I was there. I heard it. We're past all that. [laughs] That wasn't fun. My data swamps that I had to deploy. We're past all that. Actually, you know, it's funny you bring up Hadoop. 29:48 You know, I was, I was... The, the, the, the real growth hack now is becoming the tool that's suggested by the AI chatbot, right? Because people are like, "Oh, how do I build this?" 29:57 And if it suggests go use, uh, you know, um, I don't know, uh, BigQuery and, and GCP or whatnot, that's a, you know, lost customer to Hadoop, right? You know what I mean? [laughs] Yeah. 30:06 And so, like the real trick is how do you, uh, uh, how do you become the one that's suggested? 30:12 The, and so, not, not to take away from you, Dan, but, you know, given that you're sort of in the browser business, at least perceptually, you're talking about how we're getting a shrinking differentiation between OpenAI, Claude, et cetera, et cetera. 30:24 Does that reflect back on sort of the competition for, for the browser space between Netscape and Explorer and everyone else? I mean, how do you differentiate it given that, that it's hard to differentiate? 30:34 The browser space has been, you know... 30:37 And I don't wanna over trivialize this and or say this necessarily in the wrong way, but for the average consumer, they don't see a lot of differentiation with the browser anymore, right? 30:45 We're way past the sort of 2000 to 2010 era, or '90s era- Where there was big differentiation. The average consumer may not even know what browser they are using [laughs] uh, right? 30:55 And so if you look at customer acquisition in the browser category, historically it's pretty much been pre-installed, right? 31:00 If you're the platform provider, you're Android or you're, uh, Windows or you're, uh, macOS, it's installed by default, right? And that's been the sort of, um, uh, you know, that, that's been the challenge, you know. 31:10 As Firefox is a third party, how do you sort of displace that? You know, in many ways they leverage every monopolistic muscle they have, uh, to make sure that they can't be displaced. 31:19 We are seeing, for example, in Europe, uh, things like choice screens on mobile, uh, which has been great. 31:24 And, you know, it's been great for Mozilla because we do a lot of things to really protect consumer privacy, right? 31:29 Uh, and that's sort of a point of differentiator, but it's, it's a little bit nuanced and, and many don't, many don't, are not aware or many don't even care. Um, I think going forward, 31:39 you know, uh, there are some big questions like what the future of the browser is gonna look like. This is probably one of the hottest topics within just sort of the broader, uh, gen AI category. 31:48 Uh, and, and I'll give you an example. You know, if we enter this sort of ChatGPT or let's say AI chat kind of era where you ask a question and you get an answer, eventually, right now you have multi- multimodal input. 31:59 You could ask a question, you could post an image, you could post a video, but the output is almost usually text. But you can imagine that you're gonna get multimodal output. 32:06 Like, I ask a question and it generates an application on the fly that I then interact with to answer that question. 32:11 So I might ask a question like, "Hey, I wanna learn about this," and it generates a whole classroom view, uh, and whatnot, or I want to learn, you know, I want to entertain me with some cool videos about basketball. 32:21 It, it, like, generates a whole bunch of videos that are cool and entertaining about basketball, right? That's multimodal output. That's a completely different world, right? 32:30 In that case, you're no longer browsing, you're just consuming everything through some sort of aggregation interface, right? 32:35 And so I think it's, you know, there are some big open questions like what this future browser might look like, and so we don't, uh, uh, we're of course trying to get ahead of that and we're doing, uh, you know, a lot of work there within, uh, Firefox and more. 32:46 Um, but, uh, there are a lot of unknowns. Uh, uh, Raj, I was gonna ask you a similar but, uh, different, which is more about, like, I think what you're getting at, which is search, right? 32:55 Like, if, if, if you're asking a chatbot or an assistant on your phone or, you know, now there's these o- other devices now that have AI on them, right? These, like, personal assistant devices. 33:05 Um, and, and its job is to get you the best answer, um, instead of the most popular answer or the answer for whoever paid for the answer to be in front. 33:14 But at least before on a, a, a search, a search page, you have all the choices, right? To your point, and you can kind of pick which one's the best one. 33:22 How does the ch- if the AI is the one that picks the right one, does that change the business or does the business inform how we get search results? 33:30 So I mean, I think we're already seeing some large chunk of search has moved to AI chat, no doubt, right? And of course, from a Google perspective, you know, clearly the market leader here, their job is to defend. 33:42 Certainly their job would, you know, even better is if they can innovate and sort of not churn, but if they can defend, they can maintain. 33:49 Now search, very much like the browser ecosystem, um, has been, uh, a situation of defaults, right? 33:55 You know, uh, uh, and I don't wanna go dig into all of this too deeply because, you know, with the whole sort- sort of Google antitrust stuff and Mozilla and all that going on right now. 34:03 Uh, but, uh, uh, in general, you know, there's a lot of pay to play, uh, for different positions in different places and, and whatnot, and sort of how the ecosystem works. 34:12 Um, and, uh, you know, it's also a superior product, right? There's a lot of data. I mean, Google, uh, has been the best product. 34:18 Now Google has launched, uh, AI chat and, uh, uh, and, you know, it's, I think it's trending well. And if you look at a lot of the, um, uh, search market share data, I think, I don't think Google has slipped at all. 34:29 I mean, I think ChatGPT has taken some chunk. I think they've impacted some of the other, uh, search providers more, but I think Google's still hovering around 89, 90%. 34:36 So I think from a consumer perspective, yes, there might be some, you know, we've seen some behavior change, but it's not like the search companies are sitting idle, right? 34:44 They're bringing in AI chat themselves, uh, uh, and they're, uh, um, uh, you know, they're trying, th- they recognize where this is going. People want answers. People don't wanna click a bunch of links. 34:55 How that impacts the broader ad ecosystem, I mean, that's to be determined, but certainly these are things that they're thinking about. Chris, what do you think? 35:05 I mean, you know, we, we've had a bunch of shows where we talked about, I mean, there's a th- a thousand different angles to this, right? 35:10 But, uh, you know, the, uh, yeah, determining what the truth is and, and, and, you know, r- news and information, but also just, you know, uh, what's the right model in, in getting people information, right? Yeah. 35:25 Well, and, and, and you know that I'm, I'm frequently, um, doing the more cowbell drill on w- us being in a post-truth internet because how do you, to your, to your point, Raj, it's, it's, I'm no longer browsing, I'm orchestrating, I'm conducting. 35:39 I'm saying, "Create this for me," knowing that it's not real or it's part of my creation, but then does, does the rest of the universe know that? And so, you know, 35:49 I, I guess the challenge with that sort of task-based automation is does the user or the consumer that's, that's using that thing and is sort of outsourcing a degree of expertise that they, they are dependent upon, how do you confirm for them that, hey, when, when you come to me and I'm setting up, I, I don't know, a business account for your company, how do I know that I can trust the result of that? 36:08 And, and I'll give you a very specific example from a couple of weeks ago we talked about on the show. 36:12 I use a credit, an old credit union for a lot of my banking, and someone asked me if I could Venmo them, and I, I Google AI'd, "Hey, does my credit union do Venmo?" 36:20 And it's like, "Absolutely, and here's a whole list of instructions on how to get on it." It was totally fake. It made it up. 36:27 And so, you know, how do you guard yourself from that sort of thing, um, in, in leveraging the expertise that's available in task-based stuff? 36:34 But I think w- I think part of the answer, and you've touched upon it, is humans gotta stay in the loop. Yeah, no. So if you're looking like, um, I'll give you an example, like the Perplexity Comet browser, right? 36:43 Or a lot of these sort of browsers, right? They're, they're, um, uh... and where things are going, um, you wanna be able to say a task. 36:49 And weirdly, in the AI space, for whatever reason- The benchmark example everyone always provides is book me a flight. Like, [laughs] travel, I think Siri started it, but travel's always been sort of the benchmark. 37:00 A lot of people- Technology still hasn't made this easier, yeah... 37:03 But ironically, it's one of the workflows I think consumers are least likely to do, like, because they're price sensitive and they want to, like, have some control, right? 37:10 But it's always been sort of the benchmark, book me a flight or make me a reservation. Um, if you look at the user experience, there's... This is a UX problem. This is not an AI problem, right? 37:22 So like, let's say book me a flight, you're seeing the browser in front of you open up the appropriate website, go and click the appropriate links, enter in the data. It's doing the activity. 37:31 Now, it'll hit a certain point where it wants a human in the loop and it will pop up something, say, "Hey, can you confirm this sort of next step?" Right? 37:37 So I think that's how it's sort of being addressed, and the question is how that user experience, uh, evolves, but they are keeping a human in the loop. 37:44 Certainly there's been some demos where it's going end to end, and that's a problem, that's a problem, right? If we follow the 95% rule, that doesn't work, right? 37:51 Because if it books the wrong flight 5% of the time, that's going to be a big issue. So, so I think, again, I think this is more of a user experience challenge. 37:58 And of course with all of this, we should assume that over time it will get better, and a lot of it getting better may not be necessarily improving the underlying model, but doing more work around creating guardrails. 38:09 So they might have different places where they check the input and say, "Hey, you know what? We should ask for more clarification." 38:14 Or they may decide as they go along this flow that they want to no longer do it in a visual LLM way, but rather just th- now plug in directly into the API, right? 38:25 So they hit a certain step and they're like, "Oh, let me just call the TripAdvisor API through a tool interface," um, and then that way we know it's deterministic. 38:32 It's no longer, uh, you know, um, going to be non- non-deterministic because it's just hitting an API, so as long as our data's correct, uh, and the human said these are the times and whatnot, then we just book it, right? 38:42 So, so I think there's... I think a lot of this is frankly more of a UX problem. Trust but verify. Yeah. 38:49 And you know, and it's, you know, the interesting thing too, uh, you know, just sort of leaning on this, if you look at, 38:55 there's this very top, popular topic now in sort of, uh, search engine optimization, this new term called GEO, generative engine optimization, which is basically how do you optimize, appear in AI chat? 39:05 We, we alluded to this a little bit earlier. In SEO, the game has always been high authority d- high authority domain backlinks, which means get me backlinks from Forbes and CNN and, like, high authority sites, you know? 39:15 Sites that have, like, you know, high reach, right? The problem is many of those websites have AI assisted or just pure outright AI generated content, so there's a lot of slop. That's not particularly helpful to an LLM. 39:26 So SEO has been that world, but LLMs, they're looking at high authority, uh, UGC content, right? So user generated content. 39:33 So they're looking at, like, if you look at the backlinks that are most cited by LLMs, it's things like Reddit, Wikipedia, review sites like Trustpilot, um, you know, uh, comments within YouTube. 39:45 Why, why is it those places? Well, con- con, uh, comments in those places are likely human moderated. They're more likely to be human authentic, less likely to be AI slop. 39:55 They're looking at podcast transcripts, for example, like even this transcript. Why? Because this is most likely to be humans actually talking. 40:01 So it's really interesting how LLMs are prioritizing content where they feel the source is more likely to be human versus AI slop. 40:10 Uh, and so part of that is just to make sure that it lands at a su- suggestion or recommendation that is closer to the truth. Interesting. 40:21 So Raj, uh, you've, you've, as I mentioned at the beginning, you've, uh, you've started and, uh, sold a, a bunch of companies and, and I'm curious as to, you know, was there a common denominator between all of them? 40:36 Like, what, what... Was there one thing or a couple things that, you know, made them successful and, and, and sellable? Is there a common thread? Well, so first of all, nobody should aspire to sell a company, right? 40:47 So, you know, the, the game is you're p- you're playing to become a standalone business, and anytime you enter, uh, you know, the nature of the venture ecosystem when, when they invest, they see a new hot category. 40:57 Let's say five years ago that might have been remote work software, right? Today it might be, you know, uh, gen AI wrappers or whatnot, right? 41:04 So you s- you enter a new category, they now look at sort of like total assets under management. It's trillions of dollars now at this point sort of in venture. 41:11 Uh, and they might say, "Okay, we're gonna place a 1% or 2% bet in the, in the space." 41:15 That might represent tens of billions of dollars or whatnot, and, and that's fine because that's, that's really not gonna move their needle. 41:21 The whole space sort of falls apart and everyone dies because it didn't really, you know, it didn't become a category. It just sort of became an add-on and nobody stood alone. 41:29 You know, that's okay, but it's worth taking that bet for 1%, 2% of my overall, you know, AUM. Um, so uh, I think, uh, you know, I've had startups that have been in that situation. 41:40 Like, you know, like very few, uh, software products, let's say out of the remote work collaboration category made it. Mostly the big ones got bigger, right? 41:47 The Slacks got bigger, the Zooms got bigger, the Atlassians got bigger. Uh, but the startups, I mean, some emerged. You know, we had the Miro, the Murals, uh, et cetera, et cetera. Uh, right? 41:56 Um, and I've been in categories where, um, you know, uh, some players won and it became, you know, it sort of became a huge, uh, ecosystem, right? Like it became its own category. 42:05 It became its own stand, like somebody got to IPO. So the whole joke with me is I've been stuck at first base. Like, I keep getting to these like, you know, seven, eight digit exits, right? 42:13 But like, uh, uh, you know, that's, that's like, how do I get to the next sort of echelon? So I'd say that's one of the sort of core theme. I've never aspired to sell a company. Um- That's Moneyball. 42:22 That just means you're a good player. [laughs] I would say thematically, and I alluded to this, uh, at the beginning, uh, all of my startups have been consumer focused. Uh, you know, bottoms up consumer. 42:32 I, I tend to focus on my own pain points, my own problems. I, I sort of believe that to build a great product there's, uh, a few core things that are needed, like, uh, creative execution. 42:41 Like you have to be very creative. What worked yesterday doesn't work today. Like people, people are like, "Oh, you can run the same playbook." I'm like, if I was building the same business over and over again, maybe. 42:49 Uh, but no, the, the, what... It just, it, everything's dynamically changing. What worked yesterday won't work today, right? So creative execution. You need to have some product sense. 42:57 You just need to have some ability to like- Recognize what's good taste, bad taste, judgment, sensibility. That comes from reps, that comes from intuition. 43:04 But the other thing is you need to have, uh, uh, you know, really deep domain knowledge, right? 43:09 And I think that domain knowledge comes from, like, really understanding the problem, and that's what creates that so, so-called cognitive empathy, and that's why I tend to lean into personal pain points. 43:17 So I'd say the common denominator amongst everything I've built is I focus on just great consumer products. Uh, and then of course, if you build a great consumer product, that doesn't mean people will come. 43:26 But if you can layer in [laughs] uh, growth tactics and more and start driving growth, and then, you know, it ended up in sort of various outcomes. 43:34 And your, and your companies have been in, in various different, you know, solved various different problems. I- I'm curious, you know, what, what are you liking AI right now? 43:41 Any, any particular categories within AI that are kind of top of your list? I think... So a trend I'm seeing, like, people are getting very comfortable with chat-driven interfaces. 43:51 I mean, this is remarkable because this was not the case- Right... for so long, right? People have been trying AI chatbots for ages. I remember, uh, therapy, coaching AI chatbots in 2015 era, right? 44:03 Uh, and so, uh, we're finally hitting this point where people are like, "Wow, this thing's actually pretty good." Yeah. Like, "I, I trust it. I wanna use it." 44:10 And so I think the opportunity now is to really think about vibe as a modality, right? You think of the mouse, the keyboard, touchscreen, these are modalities. Vibe is a new modality, which is basically chat. 44:21 And so then the question is: how can every application that has been built to date be reinvented with a chat-driven experience, and what does that look like? Like, what would a chat-driven Salesforce CRM look like? 44:32 What would a chat-driven browser look like? What would a chat-driven, you know, uh, HR, you know, Workday application look like, right? And you could see, uh, this happening. 44:43 Many of them are trying to bolt on that experience. But the question is: what would it look like if it was built chat first, right? As opposed to sort of a bolt-on, right? 44:50 And that's kind of reinventing with the GenAI mindset. So I think you're gonna see... My, my general take is over the next few years, and what you're seeing right now, is people are so focused on the AI innovation. 45:01 I, I call it UX innovation because a lot of this is about figuring out that user experience to enable a similar experience, but through just a much simpler vibe interface. 45:11 And when you say chat, you mean text chat or could be voice chat? Either. Doesn't matter, right? I mean, voice is historically always under, um, under-delivered, right? Overpromised, under-delivered, right? 45:22 Like, beginning from the Siri era, right? There's a lot of data that supports that, like, predominant use of voice continues to be sort of at the home, um, in the car, but not out and about. 45:32 Not, you know, it's, it's because it's, it's not private, right? You don't want everyone to overhear all your discussion and whatnot, right? 45:37 But, like, uh, but now we're at a point where people are having conversation, and they're not saying it's bad. Like, that's amazing. [laughs] Um, so, um, I think, you know, uh... 45:46 And we're seeing tools like Whisper, you know, Whisper Flow, right? Like, literally on your computer where you can, where you can just, uh, turn every input field into a voice chat box. 45:54 So I can just hold down a key, I use it all the time, and I just say what I want, and then it enters it in. This is becoming more important as we're moving to vibe interfaces and everything, right? 46:02 Because I'm typing a lot more, and I don't wanna type as much. [coughs] And so I think, uh, I think it's all of it, both the text chat and voice chat. Amazing. Amazing. Well, Raj, really great conversation. 46:14 Thanks for sharing your story and, and the things that are going o- go on in Resdala. Hopefully... I, I usually tell people, "Let's, let's have you back in, like, uh, six months." Yeah. 46:22 But, you know, six months is like three years. It's like dog years now- [laughs]... with, uh, with AI. So hopefully we'll have you on again soon. All right. Thank you very much for the conversation. Enjoyed it. 46:31 Great to have you. Thank you, Raj. Interesting guy, Chris, yeah? Yeah. Well, like, uh, the Moneyball thing, it's like, you know, why, why do you want him? He gets on base, you know? That's, that's... 46:41 It's tough to be consistent from that perspective, so kudos to him. Yeah. And the, the, the idea of kinda coming from that entrepreneurial, uh, environment, though, I think also very cool. 46:50 You know, some, some of these big companies, they, you know, they still create products that are, you know, they, they think are sellable for their own, you know, corporate reasons and, and lose touch with what the consumer actually wants. 47:02 So, uh- He strikes me as a guy that doesn't-... clearly he, he's the guy that everyone wants. 47:05 He strikes me as a guy that doesn't spend a lot of time having Scrum meetings next to the latte machine as an excuse for not getting any software written, so that's good for him. That's funny. That's funny. 47:14 All right, Chris. Thanks, everybody. Thanks for joining us. If you're working on anything and you need some help, please let us know. We're here. 47:19 We're a free, uh, professional association, have a bunch of resources available. Uh, just let us know. You can go to erpaai.com, uh, erpa. uh, erpaai.com/analyst to schedule a, a briefing or a workshop. 47:33 Um, in the meantime, I'm sure the world will look very different a week from now, Chris, but I'll, I'll see you next Wednesday. All right. Next time over the rainbow. Sounds good. Take care, everybody. Have a good week. 47:44 [outro music]