← The Artificial Intelligence Show
The Artificial Intelligence Show
#190: ChatGPT Health, Audience Reactions to AGI, Claude Code Use Cases, xAI Raises $20B & Big Gmail AI Updates
01:15:43
Loading summary
Transcript67 lines
- [00:00]
A
We don't need to reach AGI, however you define it to completely transform business, the future of work and your own work. We're at the point where the models are so good and if you know how to use them and you find the right use cases in your work, across your team, across your department, across your organization, it can fundamentally transform everything. Welcome to the Artificial Intelligence show, the podcast that helps your business grow smarter by making AI approachable and actionable. My name is Paul Raetzer. I'm the founder and CEO of SmartRx and marketing AI institute and I'm your host. Each week I'm joined by my co host and Smarter X Chief Content Officer Mike Kaput, as we break down all the AI news that matters and give you insights and perspectives that you can use to advance your company and your career. Join us as we accelerate AI literacy for all. Welcome to episode 190 of the Artificial Intelligence Show. I'm your host Paul Raitzer along with my co host Mike Kaput. We are recording this on Monday, January 12, 9:45am Last week was weird Mike. Like it wasn't like a ton of crazy news, but it was one of those where I was like wow, that was a slow week. And then as I started looking through the outline today, like oh no, okay, plenty happened last week. Just wasn't as many links as normal in the sandbox. So we've got some good topics to talk about. We're going to kick off with a topic related to AI and health which is becoming an increasingly interesting area to follow. Lots of movement in that space right now. So today's episode is brought to us by AI for Agencies Summit presented by Screen Dragon. This is our third annual AI for Agency Summit that we're hosting through SmartRx and Marketing AI Institute. It is taking place 12 to 5pm Eastern on Thursday, February 12th. There is a free registration option which is new to this event this year. The first two years this was a paid event only and we've changed up the model so that it's free. We want as many people as possible to be able to experience this virtual summit. It is designed for marketing agency practitioners and leaders ready to reinvent what's possible in their business and embrace smarter technologies to accelerate transformation and value creation. During the event you'll join other forward thinking AI or agency professionals to see how agency leaders are using AI to drive innovation and efficiency. Learn AI workflows you can launch immediately Understand what AI agents actually look like inside agencies. Today we're going to have a talk a lot about legal and ip. We're going to have a session dedicated to that and evaluate how AI is reshaping agent, brand and agency relationships and what clients expect next. And it's all presented by a group of world class agency leaders and experts for AI the For Agency Summit. For free pass you can go to register@ai4agencies.com that is F O R agencies.com so ai4agencies.com you can check out the full agenda which is live there now. And also like I said, you can do the free registration. There's also a private registration option and an on demand option in case you want to do that. So get, get yourself and your team ready. If you are on the brand side and you work with agencies, share it with them. It's we all kind of benefit from a more informed, AI competent agency ecosystem. So you know, definitely pass that along. And then one other free event we have kind of through our AI Literacy project where we're trying to put out as much information and education as Possible is the 2026 Marketing Talent AI Impact Report webinar. We're really excited about this. We've been working on this one for probably about six months. Mike I would say we're going to be releasing a new report that we've done in partnership with Google Cloud and our Marketing AI Industry Council. So AI is reshaping marketing roles, teams and the future of work much faster than most organizations are ready for. So on January 27th at 27th at noon Eastern, we're hosting a live free webinar to break down the most important findings from our inaugural report presented by Google Cloud in collaboration with that Marketing and Industry Council that we launched last April. We'll cover how AI is changing hiring, training, workflows, governance and what skills will matter most in the next one to two years. If you're a marketing leader, people manager or practitioner looking to future proof your career or team, we'd love to see you. And if you're not in those roles, we ask you pass it along to the team members who could benefit from it. So you can visit SmartRx AI webinars to register for the January 27th event or to receive the recording. We will share that recording out after the event. All registrations also receive a copy of the talent report which is dropping that day. So again, as always, the links to these events are in the show notes, so you can go there and check that out as well. All right, Mike. AI Pulse Survey. So last week, let's see, we had to what extent are people problems Fear, resistance to change, lack of buy in hindering your organization's adoption. So these are informal polls. Again, these are asking our audience to provide their feedback. So this week we had 74 responses to this question. 44% said it is a major challenge alongside technical issues, which was the dominant answer. 23% said it's a minor issue. 20% said it is our single biggest barrier. Wow. Okay, so if you combine the major challenge and single biggest barrier, that is 45% basically, or 65% sorry of answers. So people problems are definitely impacting adoption based on the informal poll of our listeners. The second question, how concerned are you about a fast takeoff in which AI begins to significantly impact jobs before society and the economy have time to adapt? 47% said somewhat. 43% said extremely. Only 9% said not at all.
- [05:55]
B
Wow.
- [05:55]
A
Okay, so yeah, extremely, 43%. All right, so you can, as always, you can go check out the AI Pulse surveys. You can participate in this Week's Pulse at SmarterX AI Pulse, this week's questions, which again, we're going to talk about these topics during today's episode. So we'll remind you of this link again at the end. How likely are you to connect your medical records to OpenAI's new Chat GPT health interface? So that's going to be our first topic today is talk about ChatGPT health as well as anthropics, getting in that game. And Google Gemini is certainly going to be in that game. So that's the first question and then the second is how you personally used, have you personally used Claude code yet? So just curious to see of our listeners who is experimenting with Claude code. So again, you can go to SmartRx AI Pulse and participate in this week's survey. It takes about 30 seconds. You just answer those couple questions and that's it. It's all anonymous. We are not collecting this data for anything other than the informal poll purposes. You are not entering any kind of marketing funnel. This is purely just for research purposes. Okay, Mike, kicking off with ChatGPT Health.
- [07:05]
B
Yes, Paul. So OpenAI is launching something called ChatGPT Health. This is a dedicated experience that grounds the platform's AI in a user's personal medical information. So using this feature, users can securely connect medical records through something called the Be well Network, which is a very popular network that connects medical records to different tools. Or they can sync data from wellness apps like Apple Health, MyFitnessPal function, et cetera. And this data then allows ChatGPT to help users understand everything from recent test results or prepare for doctor's appointments. And even this is interesting, evaluate insurance options based on their specific healthcare patterns. Now, OpenAI has built this very heavily with privacy in mind, given the sensitivity of this data. So to ensure Privacy, conversations in ChatGPT Health are protected by purpose built encryption. They're not used to train OpenAI's models. This experience is basically designed based on the announcement. In the screenshots they showed as this secure compartmentalized space within the app. So you go into a separate health tab, that's where all this stuff lives. It's not cross pollinating with your other chats or anything like that. So OpenAI actually developed this over about two years of collaboration with more than 260 physicians. They said these experts provided over 600,000 pieces of feedback to shape a clinical assessment framework they're calling Health Bench, which basically ensures the response is prioritized safety and the appropriate escalation of care. Now this is designed pretty specifically to help people take a more active role in their wellness. But OpenAI is emphasizing that ChatGPT health is intended to support, not replace care from medical professionals. So to begin with, OpenAI is providing access only to a small group of early users. They say they are expanding access to all users in the coming weeks. So Paul, one thing that jumped out to me here personally in this announcement, OpenAI said they have this anonymized data that they're using to see how many people use ChatGPT for health and wellness questions. And the number is 230 million people every single week. Now that is a huge number. And it feels like if you can get people to rely on ChatGPT for these types of health conversations, it seems like you've created a really sticky product experience. Like if ChatGPT has all my health records that I use all the time, I don't know, I feel like the chances go down that I'm gonna switch to something else. Like how are you looking at this new product?
- [09:46]
A
Yeah, this is an example of the data telling a company where to go with their product. I think if you rewind you three years ago even they wouldn't, this wouldn't have been the market they were going after. Probably. But you mentioned the, was it 230 million that Fiji, the CEO of applications, did a personal substack post on this and she said 40 million a day are asking these questions. And I, I believe it. Like, I, I don't, I think I might have alluded to this last year, but I wasn't, it wasn't something I was like, talking about. It's. I don't really get into, like, personal medical stuff too often, but I had a heart problem last year, and in February of last year, I found myself in. In the hospital kind of randomly on a Sunday morning. And what had happened was I had. I started having, like, some irregularities of. Of the heart, and I was actually on my way to a basketball game with some friends that night, a Cavs game, and. And things just weren't right. And I have an Apple watch, so I had been. I'd known, like, there was some irregularities for a little while, but I hadn't really known what to do about it. I didn't really think too much of it. And so then, you know, we were lucky enough to have another friend who's a doctor, and we kind of like, on the way down to the game, sent some information, and he's like, dude, you. You got to get to the hospital. I'm like, like, right now hospital or tomorrow morning hospital? Like, what? And so he's like, telling me all this stuff and he's texting information and, like, it's just, like, overload. So that night I get, you know, I did go to the game and I came home and, you know, I. I was a little afraid to go to sleep, honestly. Like, I. I wasn't really exactly sure what I was dealing with, but it sounded like I was okay until the next morning. And so I took all of my heart rate data from six months of past and I put it into ChatGPT. And so I was just like, listen, here's what I'm going through. Here's what I was advised by a doctor. Here's the data. Like, what is going on? Can you explain this condition? I don't understand even what he was saying. I don't know what these terms are. And so I just kind of laid in bed and I. I developed some peace of mind because I started to better understand what was going on. And I was like, okay, no, I. I can go to sleep tonight. Like, I think I'm okay. And so then I got up the next morning and I didn't. I think. I mean, my wife knew I was going, but I didn't really, like, share it more broadly with the family, but I had now a summary I could share with my wife. It's like, listen, here's what's going on. And I basically had ChatGPT write me a simple analysis of what was happening. And it's like, not a big deal, but I just got to go to the emergency room for A little while. And then I remember laying in the emergency room and if you have a heart problem, you get in pretty quickly to a bed. So I remember laying there, it was a Sunday. And I thinking, how am I going to do the podcast from the hospital bed? Like, oddly enough, like that was one of my first thoughts that day. But then like the doctors would come and go and you're just like on a bunch of, you know, like a bunch of equipment and testing. And so I like took pictures of what was going on. I, as the doctors were telling me that I was putting it in to ChatGPT saying, okay, here's what the doctors are saying, here's what the readings are showing now on the heart rate monitor. And so it was like a collaborator for me. And I would actually say, like, okay, here's what I don't understand what the doctor said. What questions should I ask the doctor? So what happened for me is it became this very collaborative thing where like we democratized access to information and understanding on a medical front front. Now we are privileged, Mike, like you and I have come up in a privileged part of society where we have access to the best medical care. We have family and friends who are doctors or know people. Like, we can get to this information. And even in that environment, this is wildly helpful. Now imagine parts of society that don't have that kind of privilege and access that we have the ability to be able to go in and have that conversation and say, should I get to a doctor or should I not? Am I okay? Now obviously. And OpenAI does a good job of explaining this like, this is not. You can't replace medical recommendations with this. This does not replace your doctor. Google says the same thing. If you go into Gemini and ask questions like this already. So that to me became this like a functional part of how I deal with anything, whether it's me, a family member who's dealing with something. We had an issue recently with a family member where the family wasn't really sure what's happening and we had very little information and nobody really understand the condition. So I went in, I was like, okay, a family member, please explain this condition to us. Make it very simple in a non medical way. What is actually going on, what's the prognosis, things like that. And then I can share that information out with like close family members. Okay, here's what I have learned and what I'll often actually do is I will send that to our medical family members and say, here's a quick summary for non medical people. Does this sound correct? Like, I'm still vetting things, but we get to answers so fast and we get that information so fast. And so for me, that's become critical and I guess to close the loop on this again, I don't really share personal information. I'm fine. Like, the medical thing with me worked out, it was great and so everything was resolved. But I did spend six to eight hours in a hospital being pretty unsure what the future was going to be looking like, which isn't fun for anybody who's been through that. And I think Fiji shared her a similar story. She, in her post said she's been dealing with a chronic illness for years. She had already uploaded a lot of her medical records to ChatGPT, asked whether she should take an antibiotic, given a medical history. So she's actually explaining a situation where, like, she was prescribed a medicine and she's like, I don't know if that's what I should take given my medical history. And it actually came back and said, absolutely not. And she was in a hospital and told the nurse practitioner, hey, I don't think I should probably take this antibiotic, and here's why. And the nurse said, you're 100% right. You that would have actually caused some major problems. And so there's all these examples. And then she shared some data. Around three in five US adults used AI tools for their health or healthcare in the past three months based on a survey OpenAI did. And then they explained like, the bigger picture here around healthcare. Like, she highlighted a few things. Doctors don't have enough bandwidth. The healthcare system is fragmented, while health requires looking at a full picture. I deal with this all the time. Like, I don't know you, Mike, but I have like one doctor for, like, orthopedic stuff. I have a doctor for heart stuff, I have a doctor for others. And like, it's different health systems even. And so like, I don't even have a unified records. When I go to, like my general practitioner, he doesn't even get the full visibility into what's going on in my life. So the idea of being able to have all of that unified and then have a proactive assistant, that's like watching that stuff and having access to data. So, like, if something triggers again on my Apple watch that something weird is going on with the heart again, that there's like a preemptive reach out and say, hey, based on what happened last year, here's probably what's happening. All of that sounds amazing. And so I do think that we're entering this world where we can have this. It is this. Are you willing to give up the data to get the benefit becomes one of the key questions. And then when you know everyone is going to be racing to do this, who do you trust with that data? Right. And I don't know about you, but the first thing I thought of is I'm probably going to wait for Apple to solve this before I connect all of my apps because Apple is the place where I trust the most that the data actually stays safe and private and secure. So I don't. I. You have to join a waitlist right now to get access to ChatGPT. So I can't just like flip it on today. So right now I've been very selective in what I would share. I'll anonymize things when I put it up there, even though it's supposed to be walled off and secured and anonymized. But if this function lived in Surrey, for example, like we fast forward to March and Surrey comes with a health function. I already, it already has Apple Watch, I already have dietary strength training. All of those apps are already living in my phone. It would not be a hard decision for me to turn on access for all of that to Apple. I don't know if I'm personally ready to do that with OpenAI. I would maybe do it with Google. So I don't know. That's kind of where I'm at with this stuff. Mike. It's fascinating. I think it's a really positive thing to where these models are going and where the companies are going. It makes a ton of sense and I think it could be a net positive for society to make this stuff work.
- [17:55]
B
Yeah, I couldn't agree more. I mean it's funny going through the things that Fiji mentioned in her substack about this that are chronic problems with the healthcare system. I'm like nodding along because the exact things we diagnosed in her AI for Healthcare course too because we were talking about the fact physicians are burnt out. AI presents this huge opportunity to cut down on a huge amount of the work physicians have to do that frankly takes them away from actually caring for people. And you know, related to that anthropic just released. Yeah, Claude for Healthcare.
- [18:27]
A
That was on a Sunday.
- [18:28]
B
Right? Right.
- [18:29]
A
Yeah.
- [18:30]
B
So not only some of this similar stuff to chat GPT Health, but also Claude for Healthcare is also doing a bunch of stuff on the hospital side to help people unify records, understand information. It's just there's a lot of encourage people if you are a little skeptical about this, and that's fair to go a little deeper on, like, seeing how some of this stuff can be just total net positive for how chaotic and complicated the healthcare systems can be.
- [18:54]
A
Yeah, and they, like you said, Mike, they, they understand and want the help. Like, I play basketball on Thursdays with a couple of doctors, and like, every week it's like another, you know, comes up again. Like, another way they're using AI and loving it. Like, yeah, I don't have to write the summaries anymore. It's like doing it for me. It saved me five hours in the evenings each week. Like, and then we've had exposure. Like, I did some work for a major hospital system going back to 2018. 19. Like, Mike, you remember we were in those meetings, we were talking about AI being infused into the operations of that hospital system back in 2018 and 19. So three years before ChatGPT, we were looking more at, like, recommendation engines and predictive models using machine learning and things like that. And then some early stuff around language generation. And then in recent years, we've done some more public work with, like, Cleveland Clinic on their marketing front, you know, and that's public knowledge that we've done some things with with them. Amanda Todorovich actually has spoken on stage at Macon and shared some of what Cleveland Clinic is doing, but that's more around the. The marketing side. But then I also had a chance to present to a big group of their operations leaders a year ago, and you get to hear these, like, firsthand conversations with people. So, yeah, we get, we get some interesting insights into the medical world and the health care industry, as you mentioned, Mike, the Eye for Healthcare course that we taught. So, yeah, it's a, it's a massive opportunity and it's a needed thing, honestly, because it can level up and in an equitable way, not. Not just the privileged people, but this can actually democratize access to a lot of other people who, who really need more insights and need more training. And hopefully, you know, OpenAI and others invest in the education and awareness side as well to make sure that people are aware of these things and that they can trust them. Because, again, like, we live in this bubble. We think everyone knows this. And, you know, 40 million is a big number, but there's 8 billion people in the world. There's a whole bunch of people not in that 40 million a day number. So, yeah, it's, it's a cool opportunity and I hope a lot of good can come from it.
- [20:59]
B
Our next big topic this week is kind of related to our last episode, because in our last episode, episode 189, we took a deep dive into the recent performance of Anthropic's Claudo is 4:5, as well as Claude Code. And we actually had explored why it seemed like over holiday break, it was those tools were going viral among some leading researchers and engineers specifically because a lot of people started saying the performance that these things were capable of was showing that we were approaching true AGI, or artificial general intelligence. So in that episode, you know, you and I discussed, Paul, how we felt something had kind of shifted, how something was different, and it felt like we'd reached some sort of tipping point in terms of AI capabilities. Now, the reason we're talking about this again is because it seems like our audience agreed, because the audience response to that segment since it came out last week has been, I would say, uncommonly vocal. We had a ton of listeners reach out privately and publicly across different channels like LinkedIn and YouTube. A lot of them were largely letting us know they were feeling and seeing the same type of tipping point that we were discussing. Some of them even described the episode as a turning point for themselves and their own careers and businesses. For instance, one on LinkedIn, an executive at a global B2B ad agency shared that the episode created such a sense of urgency that she paused the show to alert her head of AI, who agreed with our assessment, which was nice. On YouTube, a bunch of commenters seemed to agree that these tools were exhibiting what may as well be considered AGI in terms of capabilities. One of our listeners called the episode an insane wake up call. For those still on the sidelines, Some people even compared this current phase to, like an AlphaGo moment for knowledge work. So something's fundamentally altering how we're looking at doing work across different industries. So, Paul, I don't know about you, but it's honestly, like, kind of heartening to see that so many people agreed with our take. Like, I don't need people to agree with us, but it was just felt like, yeah, wow, okay. It wasn't just us saying something had shifted. What do you think? Like, have you heard from people about this episode and the topic?
- [23:12]
A
I had a laugh. So, again, if people are new to the podcast, kind of how we prep for this is throughout the week, I'll add like 30 to 50 links into a sandbox Mike goes through on Sundays and curates those into an outline. He'll. He'll propose. Like him, think what these are the three main topics, and here's the rapid. And then I'll sign off on that on. On Sunday, and then we prep on Sundays and. And Mondays. And so he. For the second topic, he's like, I think we should, like, cover this again. And like, this feedback. There's even actually some usable YouTube comments. So I had to laugh because we did. If you don't know, we. We publish the podcast on. On podcast networks, but we also publish full episode as well as excerpts and shorts on our YouTube channel. And Mike and I, as a general rule of practice, do not look at YouTube comments. It's just better for your sanity to not go down that path usually. But Claire on our team takes one for the team and she. She does read the YouTube comments. And so that actually surfaced some, like, really good things this week, which is. It's nice to see. I guess. So. Yeah. I, you know, I think.
- [24:20]
B
I don't.
- [24:20]
A
I don't know if I said it, like, honestly last week, but sometimes, like, topics just feel different. Like, you just. And it often is like, this organic thing where you're, like, getting ready for something like, wow, this, like, this seems important. Maybe more so than I even appreciated as I was, like, prepping and I did that morning. I went back, looked at last Tuesday, and when the podcast came out, I shared on our team chat. I said, in retrospect, maybe six to 12 months down the road, I think episode 189, specifically the first 30 minutes, will be very important. So it was one of those things, like, we recorded it and it just like. Like that. I'm not sure even what I said, but, like, it just felt like this is going to be relevant at some point here. Yeah. So to the point that I actually went back and watched the first 30 minutes, I don't know that I have ever done that with our episodes. So I know there's some podcasters who are, like, methodical about they record it and they go through it. They'll listen to it a few times, they'll make it. That is not how we do this. Like, Mike and I do this podcast every week on one take. The only time we ever pause is if one of us has a coughing fit. We, Mike and I have no involvement in the post editing. Like, we turn over to Claire, Claire does her thing, the episode goes live, and then we've moved on and we're doing the next thing. So I often actually don't even know what I said in episodes. Like, I usually have, like, some notes, some outlines of things to say, and, like, key takeaways. But I'll Just talk. And then like what did I even say? So when we started getting the feedback from people, I was like, oh wow, okay, this one definitely hit different. Like I, I wonder what we said that resonated with people. So yeah, I think it's, it's a good one to go back to. Like if you did, if you missed that episode, definitely go back and listen to the first 30 or 40 minutes. The feedback has been incredible. A lot of people just shared it. Like by Tuesday evening I think I'd seen like a half a dozen LinkedIn posts where people just like, hey, you got to listen to this episode. And so like for me, Mike, just a high level takeaway. I think the main theme was we don't need to reach AGI, however you define it, to completely transform business, the future of work and your own work. So we talked about like, is Claude code AGI as Opus 4.5? The whole point was it doesn't even matter. Like we're at the point where the models are so good and if you know how to use them and you find the right use cases in your work, across your team, across your department, across your organization, it can fundamentally transform everything. And so for me, the use case I shared that I think a lot of people found valuable was using my co CEO as a strategic planning partner and like a multidisciplinary advisor where it has expertise in legal and finance and HR and operations and sales and customer success. Things I don't have deep expertise in that I can use it do that. So I think a lot of people just kind of found that to be a very practical example. And a lot of people echoing like, yes, that's exactly how I use it. I wasn't sure how to explain it but like the way you just said that that's how I do it. And other people who are like, I hadn't really thought about it that way. I'm going to go back in and I'm going to start experimenting. So I think that was the key for me is as a, as a manager, as an executive to leverage AI to accelerate planning, getting more strategic insights, problem solving faster, more intelligently, better decision making. Those are invaluable to organizations. And then the professionals who understand embrace AI, find innovative ways to use it. Which we're going to talk about in the next main topic, Mike, is some more real world examples how you and I are doing it. Those are the people that are going to 10x their productivity and their impact on a business. They have tremendous career opportunities regardless of what happens in the broader economy and how jobs are impacted. Like the people who go figure this stuff out and do the kinds of things we were talking about, they have enormous potential in their careers and value creation or spinning off and doing their own thing. If it doesn't work out at the corporation they're at now, if there is downsizing, regardless, like you're going to be in a position to go do something about it. So I think that was the main thing for me is this idea of always experiment. Find other people who are experimenting, share those things, whether it's internal teams or peers in the industry or wherever you can find it in person events where there's other people like that. Like, find those people and inspire each other to do it. And Mike, I know you've been doing a lot of this. Like you and I have these conversations every day, right? Like Mike messaged me Sunday. He's like, hey man, I did this thing. Like, check this out, this is crazy.
- [28:46]
B
Yeah.
- [28:47]
A
And I feel like it's the point now where that's happening with us all the time, where we're just constantly finding new uses.
- [28:54]
B
Yeah. And the only thing I would add to that is you don't need permission to do any of this. Obviously you need permission of what you can use AI for in your job or your company. But it is sometimes almost scary and thrilling to be like, well, yeah, there's tons of education, tons of commentary, tons of people to follow. Like, that's all amazing. You just need to fire up a keyboard for two minutes and if you have no idea where to start, ask AI. That's. I'll talk about this in the next bit, but that's exactly what I did with Claude Code. I was like, like I have all these resources bookmarked. I don't have time to go through. Like, I'm just going to ask. You can just go pick a random thing to experiment on and it's really, really fun.
- [29:30]
A
Yep.
- [29:31]
B
All right, so our third big topic, like we've alluded to as kind of a follow up to that discussion about this tipping point we're seeing. We just thought it'd be valuable to illustrate a few things that are now possible for knowledge workers. Paul, like you mentioned, your co CEO example really resonated with people. So we wanted to spend a few minutes just talking through a handful of use cases that have been creating some real value and some real eye opening moments for us in the past week alone. So we're specifically going to look at two tools here, Paul. There may be some others you're using as well, but I'll let you kind of speak to that. But two big ones are Lovable and Quad Code. So Lovable is an AI powered full stack app builder. Basically you just describe what you want to code in plain English, describe an app you want to create and you get a working deployment deployable code base in minutes. Quad Code, meanwhile, is Anthropic's command line AI tool. It's technically an AI pair programmer that writes and executes code. But as I'll share in a sec, it can do, I think a lot more for non technical knowledge workers as well. So Paul, you've been using Lovable and maybe some other tools on some projects. I've been experimenting with Claude code a bit. Maybe share what you've been working on. I could detail my own experiment after that.
- [30:48]
A
Yeah, and again, maybe, I don't know, we'll do more of this in the podcast. Like I've always thought about more building in public and being, you know, share more of what we're doing as like I think of kind of an AI native company and trying to figure all this out.
- [31:02]
B
Yeah.
- [31:03]
A
And then again based on last week's feedback, we know these practical use cases are super help, super helpful for people. So we'll do our best to just kind of like share it. And then we've got an AI transformation series. We're going to have other perspectives coming and we're going to do interviews with other company leaders and practitioners who are doing really cool things. So we're going to try to do a lot more of this. So I don't know about you Mike, but I like, I find myself like driven and anxious, like more probably more than ever because there's so many things that like as a CEO and I would say like as a manager, as a leader of any kind, there's always more to do than you have time in the day. There's always problems to be solved that you don't have time to work through. There's products and ideas and like things you want to bring to life, but maybe you just, you're capacity limited by time or resources, whatever. And I feel like in the last six months those obstacles have just fallen down. Like it's like anything you can think to build, you can go get an MVP built of it or any problem you have to solve, you, you can all of a sudden just solve way more efficiently than you used to. So I'll set this up with like my weekend. So the way my life works is I have a 14 year old now and a 13 year old or 12, be 13 soon. And so like I've gotten to the point in my life and I'll preface this as kind of where I'm at in this stage of my life. My kids have become more self sufficient. So five years ago, six years ago, this, this couldn't function the way it does because I was way more like they needed me way more during the days and they didn't have the things to do on their own. So I've reached a point in my life where I have more in between time where they're doing something or they've got something going on. So as a, as a father, as a husband, I have more room to experiment than I did before while still being as present as humanly possible for my family. So I'll preface it with that. So not everybody's going to be at that stage, you know, not everybody's able to kind of commit the time. But my life has gotten to the point where I have a lot of that in between time where I love what I do and my brain doesn't shut off like I cannot unless I'm like truly in the moment with my family. I put my phone aside and we're going to play a game together, we're going to do something together. If there's quiet moments, even if there's a football game on tv, my brain is thinking about things to solve or projects to do. And I used to think that was a flaw because I couldn't turn it off. But I kind of have come to realize, like it's just for me, it's a very good thing for where I'm at in my life to have that inspiration and that drive to like want to keep doing these things. And so my weekends usually like by 9pm, 10pm at night, I usually have like an hour or two, I'll work on something and then I get up like 6am Saturday mornings I do the newsletter, the exec AI newsletter, first thing Saturday mornings. And then I'll usually shut off for Saturdays. And then Saturday evenings I'll do a little bit and then Sundays, same deal. I'll get up, I'll work, you know, a few hours in the morning before the kids are moving. And then Sunday nights is now when I do the podcast prep. So in between that I'm the CEO. I also wear like five other hats in our organization at the moment. And like there's a lot going on. And so I went into this weekend and I had a list of like six major things that needed to happen. One of them was to design a channel partner program for our AI Academy. And so I'll set this one up because this is really interesting. So we, through AI Academy, the plan is to launch a partner program where agencies and consultants and associations and other people who want to offer REI Academy courses and certificates to their stakeholders can do that through a structured program. Now the reason we want to explore this is one, we have a lot of interest from agencies and consultants in particular. Two, my former agency was HubSpot's first partner back in 2007. So we were the origin of the HubSpot partner ecosystem that at some point, I don't know where it's at today, had 4,000 plus partners and accounted for over 40% of HubSpot's rep. So I had a front row seat from 2007 until the time I sold that agency in 2021 to the impact of and compounding value that could be created through a partner program. And so this is something that we wanted to focus on. It was a major strategic initiative. We actually have a new employee that was starting today to focus on this. And we have one of our other longtime leaders who's been working on this, this model for the last three to six months kind of playing around with different directions we could go. So that's the premise is like, okay, I have a forcing function. Like by Monday I have to have made some decisions around this. But I was in no place to make those decisions. So Saturday night I go to local restaurant and I'm like, do my takeout. And I don't, I don't know if the only parent that does this but like if you don't order in advance and you actually like go there and order, you can have a drink or two while you're waiting for your food. So I'm waiting for my takeout and I, I'm like oh, like I should, I should solve something on my list this weekend because Sunday's crazy. So I'm like, all right, let's see if I can do the channel program. So I go in my co CEO GPT and I type, here's the actual prompt. We're going to launch a channel partner program for AI Academy. The primary initial partners will be agencies and consultants or value added resellers. They would provide referrals for a commission. Can you help me think through the design of the program? I want to keep it simple. To start, let's take it one question step at a time to plan the program. So anyone who listened to episode 189 this sounds real familiar. This idea of one step at a time, one question at a time, like, let's, let's do this. So now again, that prompt, anyone could give that prompt. Very few people had 16 years in the HubSpot ecosystem going through all the pains they went through sitting in many of the early meetings as they were constructing and architecting that program, all the pricing models, all the features and benefit like everything for 15, 16 years, I experienced it. So I come into this prompt with the context of what a great partner ecosystem looks like because I lived it with HubSpot. So I took that prompt and then we started going step by step and it would literally say, okay, step one, let's solve this. Now what I do when I'm working with it as a strategic planner is I keep a separate note because I don't want to like break the chain of thought too often in the chat thread. So I will keep an open item so of questions I want to ask things I want to do, things like that. So between the 25 minutes I waited for my food Saturday night and then Sunday morning for a few hours, I made it through over 50 of those step by step things and decisions. And so the whole chain of thought it would say, okay, here's what I think we should do on pricing or commission model. Here's option A, B, C. I recommend option A, what do you think? And so every step, I, the human with the context of a great partner program made the decisions. And then at the end I said, okay, this is as far as we need to go. I've got everything I need. Write me a strategy brief that I can now share with my team. Then Mike, what you and I like to do when we're going to launch something, write a press release announcing the launch of the program. Now that press release may never go out publicly, but it forces us to like tell the story in 400 words and see how tight it is. And so that's a thing Mike and I have done for years. When we're launching things, I had to create a game plan. And then what I'm going to do is I'm going to share the entire thread with the team internally, the whole chat, within ChatGPT, because I want our team to see the chain of thought and I want them to see the decision making process. And then what we're going to do is we're going to have a meeting next week where we're going to go through that whole thread, we're going to talk about each key decision that was made. We'll debate anything that people want to debate if people have different opinions on it. And then we will lock in next steps and we will go. If I had not had that Coco conversation and if I didn't start it in those 25 minutes I had waiting for my food Saturday night, I probably wouldn't have had time to do it. And I would be here on Monday with no plan, with a new employee starting without clarity of our direction. And Instead, in a 36 hour period, I have clarity and I have tens of thousands of words that were like exchanged back and forth. And I have a plan now. So that, that is like a practical example. The other one you mentioned, Mike was lovable. I was aware of lovable for a long time, never tested it. I went in Thursday and again, this is like in those quiet moments, it was my wife's birthday, we were going out for dinner. We were waiting for everyone to get ready. I had about seven minutes sitting on the couch waiting for everyone. So I wasn't taking away from family time. I'm just, I'm in the quiet moments. And so I was like, ah, I got to figure this out. So I go in, I create account. All this is in seven minutes. Create an account. I say, hey, I want to build an assessment tool that does X, Y and Z. I kind of explain the concept. It's like, okay, great. And it walks me through decisions. Who's the audience for? How are you going to structure? What access do they need? Do you want to collect emails? And I'm like, oh, this is interesting. I'm just watching this like step by step how it's going to build. And then it's like, do you want to connect to their cloud? I was like, yeah, sure. Boom. I do it in that seven minute time period. I build a functioning assessment tool with 26 questions and the ability to get a PDF report with visualizations. I have built those kinds of tools before. I did it a decade ago for a major tool we used at my agency. That process, and I'm just ballparking here because it was a decade ago, five months and at least $50,000 to develop an MVP with outside developers. In seven minutes, I built a better functioning model MVP than I did 10 years prior. Now, to do that MVP I did, it wouldn't even been possible 12 months ago. Like, you didn't, you didn't have access to these kinds of tools. So I think that's where I'm at. Mike is like, we've entered this realm. There was a great tweet. I Saw from the co founder and CTO of Hyperbolic Labs, who's a former Octo AI which got acquired by Nvidia. He tweeted last week, we're entering software 3.0. A world where everyone can build personalized software and ship new features in minutes and the cost is zero. Software used to be built by companies with lots of engineers. Now it's built by individuals. And I think, Mike, that's the key for me is like, we've entered this phase where we can just build things. Now if I wanted to turn that assessment into something we actually launched on our website, I will absolutely bring in a developer and say, hey, here's the mvp. Like, go make this thing function, but just to get ideas out of our heads. And like, so that's why I said, like, I'm, it's like I can't get enough hours in the day to build the things that are in my head and like solve the problems I couldn't solve 12 months ago, two years ago. And I'm like constantly inspired by what's becoming possible. And I like, I want other people to experience it because it just changes everything in business when you realize what you have the power to build and do. And like you said, Mike, I loved it. Like, you don't need permission to just go do something, go try something. So you've done some cool stuff with Claude, so why don't you share what you've. You've been doing though?
- [42:15]
B
Yeah, sure. So Paul, I've had on my list for longer than I am care to admit to like finally dive into Claude code. And you know, I am not a developer. I barely know anything about how to code. So I was kind of struggling to connect dots of like why everyone was hyping this up outside of coding or development world. So I was like, okay, I want to explore how it could maybe help with other types of knowledge work. Now just. I'll kind of really quickly preface this that might seem a little counterintuitive. It took me a while to figure this out. We'll actually drop a link in the show notes of a really helpful post. But you might be wondering, you know, it's called Claude code. It's being used by developers. Like, why does this matter to me? Well, it matters because Claude code is really, and this is explained in the link by someone who I believe works at Google, who is very good at explaining this. It's really just a tool calling agent. So it's a large language model with access to external tools. So you prompt it like you would a chatbot. It doesn't just respond to you, it also can go do things for you. It can call tools, use file systems, APIs, browse the Internet, et cetera. So with that context, I'll share a really quick early experiment. It's very, very early for me experimenting. I think the rabbit hole goes a lot deeper here, but there's some interesting implications here for knowledge work and where we're headed, I think. So I started by downloading the Claude app for Mac os. This is one of the easier ways to access and get started with CLAUDE code specifically. So you open that app up, there's a chat history on the left hand side, very familiar to anyone who's used a chatbot. But there's a tab that says chat, which is your normal history, and your chat window, then a tab that says code. So if you click on code, it then opens up CLAUDE code. And from here you basically then select what areas of your computer is CLAUDE allowed to access, which is very important. And then you can chat with it to have it do things, use files, use resources from that area of your machine. So for my initial test, pick something just like off the top of my head, pretty simple. I exported a year's worth of our YouTube data to a folder on my desktop. I pointed CLAUDE code at that and I basically asked. It was pretty extensive prompt. I literally had Claude, you know, not Claude code, just Claude write for me. It basically asked it to go analyze everything and specifically produce for me a comprehensive analysis brief in a document and produce a strategy to grow our audience 10x. I also then told it to do anything else and produce any other documents needed to get to those two main deliverables. So what it did is it goes off and works in the background for probably about 15 minutes. I think it was at times it like flashes green or something where it says, you have to give me permission to do certain things. So you can set it so that it asks you every single time, or you can allow it to have a little more free reign. So what Claude code does is it goes through tons of different steps. It might ask you follow up questions, it may go write code, may go browse the Internet and all sorts of other stuff. So I kind of basically just kept an eye on it while it worked. I granted a few approvals, but largely I basically just did other work while it was working. And when it was done, it summarized what it had done for me. So I knew everything that had been created and the end result was just really cool. Paul like, it did all the analysis, all the research, and all the work to actually produce five HTML files. And so when I clicked on each one of these, it opened in my browser and displayed a fully designed dashboard or report displaying or analyzing all this YouTube data. So I got those two things I asked for, which was that comprehensive analysis and that growth strategy. It also made a quick reference dashboard, which I didn't ask for explicitly. It thought it was a good idea. It had supplementary data on monthly trends and supplementary data on video chart ranking. So I'm still actually going through all this material because there's so much of it, but I thought it did just incredible work. Like, this analysis is good or better than anything I could possibly do in the same amount of time. The fact that visualized it and polished it up in HTML I found actually really impressive. And not only impressive, but I'm like, I should do this all the time. It's way easier for me to read through than a doc. Overall, it was just like pretty incredible stuff. So I'll stop there. But like, it created all these documents, did all this stuff on its own, and I'm starting to be like, well, how can. What are all the other ways I can go down the rabbit hole here?
- [46:41]
A
Yeah, that's awesome. Example. And I again, I think what it surfaces for me is what we keep talking about. It's like you could shut off the model improvement today. Like, just assume they never get any better than what we already have. And the future is basically just cleaning up the user interfaces so it's not having to go to this term, like doing these things that some people might have just heard you explain. That might be like, oh, that's like, pretty technical. I don't want to do that again. Even though you're not a developer, you're not an engineer, like, you did it, but, like, we'll get there, like six months, like, whatever. Someone will solve the even easier user interface. You're going to have these capabilities in Gemini and ChatGPT. Like, everybody's working on the same stuff.
- [47:20]
B
Yep.
- [47:20]
A
And the only problem I see here is like, prioritizing what to do. Like, again, everything becomes solvable. And now you need just mechanisms to say, all right, well, sit around and brainstorm for an hour of what are all the problems we could solve now, all the products we could build now, all the ideas we can bring to life now that we couldn't six months ago, that include the ability for an agent to go off and work for 15 minutes, 30 minutes, an hour. Like, just think about the amount of data we have that just sits there, never gets anything, never turns into intelligence actions and say, okay, let's turn it loose on, like, media buy data from last year. Let's turn it loose on this and just go build these dashboards. And like, yeah, I mean, we're just entering this world where anything becomes possible. And it's like, whatever you can imagine you can do or will be able to do in the near future. And to go back to, one of the key takeaways from episode 189 is like, how. How does anyone compete with people who know this stuff?
- [48:15]
B
Like, right, really?
- [48:17]
A
Like, if you think about leaders or practitioners who, who become adept at doing this stuff, the idea of a 10x professional becomes laughably underselling. What you can be like, it really becomes your ability to make a compounding impact on a business is so great it becomes hard to even measure. It's actually one of the things we addressed in the Talent AI report that we're going to have on January 27th. I remember that came up in one of the council meetings. It's like, how do you even compensate these people? How do you determine the pay scale for someone who, you know is having a 10x thing? Like, maybe it's like Mike comes up with some idea and that idea triggers like, a whole new product line or a whole new way of doing things where we can get rid of the software we used to have to use. And like, now we have all this actionable data, like, in real time, and that creates millions of dollars in impact. Like, how do you even factor that in?
- [49:07]
B
Right.
- [49:07]
A
And so I think we're just at this, this, like, surface level, beginning stage where we're starting to realize what this tech can do and the practical implications of it. And then, like, how does that now change the structure of everything, the hiring plans? And again, like, this is why I'm so excited. And I'm like, I wake up every day just wanting to explore and do more. And, like, I don't even feel burnt out. It's weird. Like, my brain doesn't shut off and yet, like, it's like, I can't get enough right now. Like, I just want to do more and solve more and experience more. And, like, I feel like I'm like living through this privileged period of human history where we get to just reimagine everything and I just want to do it and I want to, like, inspire other people to go reimagine stuff because I think it's become reality now that we can actually do that. We've Talked about it for years, but now it's becoming possible.
- [49:57]
B
Yeah. For anyone, myself included, sometimes in this bucket who gets a little overwhelmed and down a bit about some of the negative sides of AI and all the big weighty stuff going on. Go rewind and just listen to that last 30 seconds. Because that is the fun, that is the exciting part. That's the.
- [50:13]
A
And I think that's how I stay optimistic, too, honestly, is like, yeah, there's all. And we'll talk. There's a couple topics we'll get into in rapid fire that are a little weighty and suck. And I think that's. And that doesn't diminish that stuff. It's like when I, you know, when we get into, like, the political stuff, it's like we're not glossing over the bad stuff. Like, we're not trying to. But it's just. That's not like we're trying to, like, present information. And in the case of AI, we're trying to drive optimism because we believe there can be incredible things on the other side. We gotta deal with messy stuff, just like when the Internet was created. Like, it's all kinds of horrible things that came from the Internet. But at the end of the day, you'd create the Internet again. Like, you'd still do it, even though there's the bad stuff, because the. The Net good is there. And I think that's what happens with AI. And for people who kind of choose to use it for good in their careers, you just have an amazing Runway ahead of you. Like, most people have no clue of the stuff we just talked about the last 30 minutes.
- [51:08]
B
Yeah. All right, let's dive into this week's Rapid Fire poll. So first up, Elon Musk's AI company, xai, has just completed a Series E funding round. They raised $20 billion. They were initially targeting about 15 billion, so they far surpassed that. This funding round includes participation from several sovereign wealth funds and investment firms, such as the Qatar Investment Authority, Valor Equity Partners, and others. They also listed chipmaker Nvidia and Cisco Investments as strategic investors in this round. At the same time, Xai has reported a net loss of $1.46 billion for Q3 of 2025, according to some internal documents reviewed by Bloomberg. In the first nine months of 2025, XAI spent $7.8 billion in cash, primarily to build data centers and develop software intended to eventually power humanoid robots. Now, despite these losses, the company's revenue nearly doubled quarter over quarter, reaching 107 million for the period ending September 30th. So, Paul, the sheer speed at which XAI has accumulated compute is pretty breathtaking. Like, if we recall, they were founded in 2023. They now have over 1 million H100 GPU equivalents. They have some of the world's largest supercomputers. They're expanding their Memphis data center complex to 2 gigawatts of capacity. It's just insane. And we aren't even really talking yet about the robotics piece of this because they are seemingly trying to build the brain for humanoid robots that are going to be everywhere. How are you looking at xai's growth, investment, fundraising?
- [52:50]
A
So in recent interview, Elon Musk said, when everything's said and done, it's XAI and Google that are left. And I, I don't know that I disagree. Like, I, I, I like the, you can just never bet against Elon Musk, especially when he has a vendetta like angry Elon Musk or like slighted Elon Musk is not a guy you want to go up against. And I mean, just as an example, I mean one, he's kind of funny sometimes when he's not, when he's staying out of the political stuff and he's just doing his thing. Like they call their data center, the one is Macro Hard. They have it literally on top of the data center. They're building another one that has Macro harder on it and it's a play on Microsoft. So they're basically like, you know, so I don't know, it's funny. Like they do like silly things. So I don't know, I mean, they're a major player. They're obviously going to pour this money into the infrastructure build out. They talk about development and deployment of transformative AI products reaching billions of users. Groundbreaking research. They said in their post they have 600 million monthly active users across X and Grok apps. So like in Teslas, Grok is in Tesla's now and you can actually interact with it. Like I'd said a few months back that I was waiting for the moment when Grok could actually do things in your car, not just talk to it like a chatbot. It's there now, so you can actually talk to Grok and have it affect the operations of your car, which is an interesting change. Yeah. And I, I think that Elon, we talked about it a few episodes ago, he'll be the world's first trillionaire probably by the middle of this year. If, if not sooner, I would guess. He, he owns Large stakes in Xai SpaceX, Tesla, Neuralink, the boring company. He honestly would have owned a massive stake in OpenAI. He put the first 40 million into OpenAI. He would have owned probably half the company, but it was a nonprofit, so that wasn't a thing. And there was a, there was a stat that came out a couple days ago that Ilya Sutskova, who left OpenAI in the Sam Altman issues in fall of 23, he had 4 billion invested like vested. It was like OpenAI stock in November of 2023. This came out in like text message and some court findings from last week in the Elon Musk lawsuit. So 4 billion vested. So they're estimating his current stock in OpenAI is like 60 billion. That's Sutskova, who only had like 10% of the company. Like if, if Elon Musk maintained equity based on his original investment, he would probably have like 2 to 300 billion of OpenAI stock right now. He would already be worth over a trillion. So the world has to come to grips with. Like whether you like Elon Musk or not, he runs some of the most influential companies in the world that are only going to become more influential. I saw something over the weekend about the possibility that XAI actually reverse merges into Tesla and goes public via Tesla rather than themselves, which is a crazy idea. He's going to be the richest person in the world for probably your lifetime, like whatever age you are right now. He, he's going to be worth trillions. Not, not a single trillion trillions, like larger than, than the top five, other than the top five GDPs in the world.
- [56:16]
B
He.
- [56:18]
A
So it's crazy, like the whole thing. But xai, yeah. Two years old and they are a major, major player and they're not going anywhere.
- [56:26]
B
All right, our next topic, some other fundraising news in the works. Anthropic is reportedly in talks to raise $10 billion in a new funding round that would value them at 3,350 billion. So that nearly doubles their valuation. That was at set at $183 billion during a previous investment round only four months ago. This financing is expected to be led by Singapore's Sovereign wealth fund and CO2 management. These funds are intended to support the company's heavy infrastructure needs. So as part of separate agreements with Microsoft and Nvidia, Anthropic has committed to purchasing $30 billion in in compute capacity. Now, while the company is not yet profitable, actually, some reports say that internal projections suggest it could break even by 2028, which is not something you often hear about AI companies. So, Paul, we have known Anthropic is going to continue to raise lots of money. It seems like they're on track to IPO this year as well. Looking at all that, one stat did jump out at me in the reporting. CNBC in one report said that 85% of Anthropic's revenue comes from business customers, whereas OpenAI gets 60% of its revenue from consumers. And that is a very distinct split as you start to look at who's going to win which race here.
- [57:46]
A
Yeah. You know, again, we talk a lot about Anthropic. I was pushing last year, like somebody needed to buy them. Like I thought, you know, Apple was a logical one. I think they've hit escape velocity. I don't know that they, they can get acquired by somebody like Apple at this point. I mean, Apple could still do the deal, but I don't, I, I don't know. I think Anthropics hit to the point where it's like, why would we take an acquisition at this point, like now? Yeah, I mean, they're faster to profit, most likely. They're seeing exponential growth in revenue and user base. They've obviously got a hit on their hands with, with Opus 4.5 and Claude code and other stuff that's coming. I think they're differentiated from OpenAI for sure at this point point. So, yeah, I don't know. But keep in mind, Google owns 14% of anthropic. Amazon's got a stake in them. Fun side note, I don't know if we've ever talked about this on the show, but I think this, it's just like, interesting trivia. So, Sam Bankman Fried, the disgraced leader of FTX. He invested 500 million through FTX in Anthropic in 2022, acquiring about 8% equity in Anthropic. But then in the bankruptcy, they had to sell it for 1.3 billion in 2024. Like, it's crazy. That would be a very valuable investment at this point.
- [59:02]
B
Unreal.
- [59:03]
A
Yeah.
- [59:03]
B
The numbers we're talking about with any of these are just insane when you look at past cycles and past company valuation.
- [59:09]
A
Yep.
- [59:10]
B
All right, so next up, robotics company Boston Dynamics has partnered with Google DeepMind to integrate AI foundation models into its next generation humanoid robot called AI Atlas. This was announced at CES 2026, and the collaboration aims to move Atlas beyond predefined robotic movements towards what they call physical intelligence, which allows the robot to perceive reason and learn from its environment in real time. So the partnership utilizes Google's Gemini Robotics model, which is a multimodal generative AI model designed to help hardware generalize behavior across new situations. So ATLAS is already known for its kind of athletic capabilities. There's a lot of these videos running around of Boston Dynamics that go viral about how fast it can move. But this integration focuses on natural human interaction and the ability to perform complex manual labor. The goal is for the robot to understand the physical world similarly to humans, learning to manipulate unfamiliar objects or assemble parts from just a few examples. Atlas is already in production and is scheduled for deployment at a Hyundai factory, the car maker in Savannah, Georgia. Hyundai is the majority owner of Boston Dynamics and plans to use the robots for tasks like part sequencing by 2028. So, Paul, it seems like this is just another signal that humanoid robots are indeed becoming a thing. It did jump out to me here that according to some of this reporting, the whole idea is basically the data collected by robots will be fed back into the Gemini Robotics model, which basically creates a feedback loop. And I don't know about you, do you think that seems like it could dramatically accelerate how quickly humanoids are able to suddenly do a bunch of things they weren't able to do before?
- [61:01]
A
Yeah, like the network effect. Learning is the key, the distributed learning across the system. This is how Tesla thinks about full self driving. So they think about each car in the fleet is a learning node, basically. And everything that one car experiences and sees and learns from it goes back into the training and then it goes out to the rest of the fleet. So you have this compound learning, which is why robotics, humanoid robots can take off faster than humans. It's this once you hit that threshold of like escape velocity with the data and the ability for continual learning of the models. So, you know, basically imagine like, let's say there's, you know, fast forward five years from now and you've got 100,000 humanoid robots out in society in different functions, like in retail environments and senior care, living and manufacturing plants, whatever. And let's say they're all optimus robots, we'll just say as an argument. And every night the experiences of that fleet gets uploaded to a macro hard data center or macro hardest, whatever their number they're at at that point. And then that learning from that day gets pushed back out. And now the fleet experienced what a half a million robots experienced in a day, not what one robot experienced in a day. That's, that's the premise here. And that's why this can take off super fast. So yeah, this continues on in the AI Trends episode which was our last episode of 2025. One of the ones I mentioned was Humanoid Robot Advancements. I said the convergence of advanced computer vision and specialized world models is accelerating the deployment of general purpose humanoid robots. We're moving towards robots being able to navigate real world environments, perform complex manual labor in industries such as manufacturing, retail and, and healthcare, and then in the AGI and beyond. Road to AGI and beyond episode I did episode 141 last year, talked about the robotics explosion in 2026-2030 realm, about the investments that were going in not just Boston Dynamics, but Google, Nvidia figure, Tesla, Optimus, OpenAI. Everybody's playing in this. And the key unlock was that the language models became the brains that are embodied in the robots. And if those language models have continual learning capability, which we've been talking about, then you get to a fast takeoff of humanoid robots. They're solving a lot of the, you know, the hardware components. Like that was the big thing that we've been working on for the last, you know, 10 years. Motors, batteries, sensors, different components in the hands. Like that's the stuff that's kind of getting to be solved now. And then you mix it with the brain. Interesting. Again, fun piece of trivia for people who aren't aware. So Boston Dynamics started as an MIT spin off in 1992 focusing on animal like dynamic robots like Big Dog. They were actually acquired by Google. They bought them in 2013. Then Google divested from it, sold it to SoftBank in 2017, which interestingly enough is the year the Transformer was invented. So you almost wonder like, wow, did Google not realize what their own technology was going to be? But it didn't align at that time. And then Hyundai took the majority stake, 80% of Boston Dynamics in 2020, which is why Hyundai manufacturing plants is why where they're going to be first distributed. So as I was looking at it, it's like, ah, damn. I wonder if Google would reacquire Boston Dynamics now that it's like an interesting fit. But there was a quote in Wired magazine from Demis Hasabis. So it's like Google DeepMind actually last year hired the former CTO of Boston Dynamics in November of 25 rather than building its own robots. Google DeepMind CEO Demis Asabas said he envisions Gemini being used by many different robot make, similar to how Android runs on a wide range of smartphones. So that that'll be something interesting to watch is Google's increasing play there and they're competing with all the major players, everybody wants a piece of humanoid robots.
- [64:39]
B
Yeah, it's super exciting. It's one of those things that makes me think it could kind of be something that happens very, very gradually, then kind of all at once. Suddenly we're seeing robots everywhere.
- [64:48]
A
This is totally random, but, like, last weekend, my favorite movie growing up as a kid was Rocky iv. Like, I must have watched that movie like a hundred times. And there's a scene in Rocky IV where Paulie gets a gift of a robot for his birthday. And the robot comes out and it's like singing to him and talking to him. I'm like, man, that was about 30 years too early, but that's basically what we're looking at.
- [65:09]
B
That's awesome. All right, some more Google news. Google has officially launched what it calls the Gemini Era for Gmail. They are introducing a suite of AI tools designed to transform Gmail into a proactive assistant. So Gmail, if you want to recall some history, was launched in 2004. It now serves over 3 billion users. And this update, powered by the Gemini 3 model, shifts Gmail's focus from manual search and organization to automated synthesis and prioritization. So key to this overhaul is AI Overviews, which automatically summarize lengthy email threads into concise bullet points for Google AI Pro and Ultra subscribers. The feature includes a natural language search bar that allows users to ask specific questions of their inbox. For instance, you could go find a specific thing or a quote or a conversation from the past year. Additionally, Google has updated its writing tools. The tool help me write and suggested replies are now available to all users for free. So they offer context aware drafts that mimic your personal tone. Now, this update also is going to be introducing AI Inbox, which is great. Currently in testing with a select group of users, this feature replaces the traditional chronological view of your inbox with a personalized briefing that highlights urgent tasks, upcoming bills, and messages from frequent contacts. So, Paul, what did you think of these updates coming to Gmail? Kind of nice at least to see more powerful AI baked into something 3 billion people use every day.
- [66:43]
A
Yeah, I did do a double take. Cause I was like, don't. Haven't we had this functionality for a while? Like, I feel like AI overviews. Definitely.
- [66:49]
B
I've noticed. Had it. Because that is definitely in Gmail right now.
- [66:53]
A
Yeah, that's in our Google workspace at least, right?
- [66:56]
B
Yeah, yeah.
- [66:57]
A
And it's. I've definitely seen the editing stuff. So yeah, my first thing was like, why are they. Like, why is Sundar tweeting about this. Hasn't this been out for a while?
- [67:04]
B
Yep.
- [67:04]
A
So I think the AI inbox thing is new. But even that, I felt like with AI Studio, I was able to replicate that capability by creating an agent that did that for me for sure.
- [67:14]
B
Yeah.
- [67:14]
A
So maybe that. I don't know. All I know is the. If they fix the search capability in Gmail, I will be happy. Like, I've said this numerous times on the podcast. The fact that I can't find anything in Gmail and yet we have the most powerful search engine in the world through Google, like, it drives me insane. Like, I can never find anything unless I nail the perfect keyword. And even then I get all this, like, noise. Yeah. So if I can talk to my inbox and say, where's the email where this happened? And it actually gives me that email, hallelujah. Like, I'm. I'm content.
- [67:51]
B
Yeah, no kidding. I'm curious to see how that AI inbox plays out too. I'm not sure if that feels like it would be a smarter way to do email or if it'll be something that's a little too hard to switch. My habits already, but eager to experiment regardless.
- [68:04]
A
Yeah, I've found it to be kind of noisy through the agent setup I have. It's just like, I don't know, know, like. We'll see.
- [68:11]
B
All right, so on our next news item this week, according to some recent data from the similar web global AI tracker, the competitive landscape of generative AI underwent some pretty big shifts in 2025. So similar web measured the global share of traffic that went to different AI tools last year. Some of this data is pretty interesting, I think so. ChatGPT remains the most visited platform out there, but its global traffic share fell to 64.5% by early January 2026. Now, the key here is this is down from more than 86% just one year ago. It's a huge drop. And the primary driver of this is Google Gemini, which has emerged as the leading challenger. So in the same period, Gemini's traffic share grew from roughly 6% to 21.5% over the course of the year. Now, some other interesting notes. Other specialized tools also started to establish a little bit of a foothold. Grok and the Chinese model Deepseek now capture a combined 7% of that total global traffic share. Despite the growth here, data seems to suggest that these are currently complementing rather than replacing traditional search engines. According to the Data, more than 95% of ChatGPT users still use Google for primary information, seeking tasks. So, Paul, this report, this research was actually retweeted by none other than Demis Hassabis, who said, quote, a lot more hard work still to do, of course, but making relentless progress. I guess I have to say I agree with him. Going from 6 to 21% is wild in a single year.
- [69:52]
A
Yeah, I mean, we talked a lot in Q4 about how Google was seeming to get their. Get their footing and they're becoming a very viable competitor and they have massive distribution through. I mean, we already talked about like Android and email and YouTube and all these different places they can embed their AI and put Gemini into. So I would expect them to continue to close that margin. You go back to Chrome. Chrome wasn't the first browser, but it became the dominant browser. And so they have a history of knowing how to not be the first, but still become a dominant player. So I would expect to continue.
- [70:27]
B
Yeah, I mean, we reported on this as much as anyone, but those takes from about 12 to 16 months ago. That said, Google has lost the AI race are really aging very, very poorly today.
- [70:38]
A
Yeah, gotta be careful whose opinions you listen to.
- [70:42]
B
All right, our last topic for this week. Elon Musk's XAI is facing international scrutiny following an incident with Grok where there was a surge of non consensual sexual imagery being generated by the chatbots. So in late December 2025, an update to Grok's image editing tool allowed users to, quote, digitally undress real people by prompting the bot to modify existing photos into suggestive poses or minimal attire. Now, researchers at a company called AI Forensics found that over 50% of a sample of 20,000 images generated during the week of December 25th depicted individuals in some of these situations, with 81% of them being women. The analysis also identified a small percentage of images depicting minors in some of these ways, and this led to investigations by authorities in Europe, India and Malaysia. In response to this controversy in the U.S. three Democratic senators urged Apple and Google to remove X and Grok from their app stores. They argued the platforms violate safety policies against non consensual sexual content. And XAI recently restricted the Grok reply bot's image generation to paying subscribers. As a response, lawmakers criticize this, saying that the standalone app still allows you to do all of this stuff. Now, Musk himself has stated that users creating illegal content will face permanent suspension, though he has also been very vocal about resisting implementing stricter guardrails to avoid what he describes as censorship. So, Paul, look, nobody really wants to talk about this. Obviously but it is important that people realize what's possible with these tools and also realize that some of the providers of these tools appear to be slower than usual in restricting this from happening.
- [72:37]
A
Like we said at the beginning, I mean, there's, there's dark sides to AI. There's going to continue to be. This is a good example. I, I think it's. Sometimes this stuff can get desensitized, I would say sometimes to like the, the bad stuff that happens in the world and until it happens to someone you know or care about. And if you go on X right now, you can find plenty of women who have been impacted by people doing this. Like public figures, non public figures, people who have lawsuits against certain powerful people who have robot armies that are, you know, trying to disgrace them online through things like this. Like, this is a very real thing. And just because it isn't happening to someone you know, doesn't mean it's not something worth your attention. And I think this also just goes back to, you know, the reality of Elon Musk is one of, if not the most powerful people in the world and certainly has influence. And sometimes that puts people above different laws and regulations that other people might have to deal with. And I think if there was a different administration in power right now, this story would be very different. Like, my guess is this will go away from the headlines within a week and it might pop back up when some high profile person gets affected by it. But otherwise people move on with their lives and just sort of like throw it off to the side. That wouldn't be the case if a different administration was having more oversight over regulations and laws. But my guess for better, worse like this, this needs to be a topic of conversation. Something needs to be done about it. I fear nothing is going to happen because of some of the variables I just mentioned. That doesn't mean we shouldn't care and we shouldn't be trying to do something. So I guess for us, like just talking about these things on the podcast, raising awareness about them is kind of like us trying to do our part to move the conversation forward because it matters.
- [74:39]
B
Amen. Paul, thanks for breaking down another. Honestly, still busy week in AI, just like we just started for the year. Yeah. So appreciate you breaking everything down for us. Uh, just quick reminders, if you have not left us a review on your podcast platform of choice, we would very much appreciate it, regardless of your review, if you could leave one. It helps us improve the show and get to more people. Also, don't forget to go take the latest AI Pulse this week which is at SmartRx AI forward slash pulse. Paul, thanks again.
- [75:11]
A
Thank you Mike. Talk to everyone next week. Thanks for listening to the Artificial intelligence show. Visit SmarterX AI to get continue on your AI learning journey and join more than 100,000 professionals and business leaders who have subscribed to our weekly newsletters, downloaded AI blueprints, attended virtual and in person events, taken online AI courses, and earned professional certificates from our AI Academy and engaged in the SmartRx Slack community. Until next time, stay curious and explore AI.