AI Time Machine, Dario Amodei Interview, and Job Role Flexibility

Welcome to wake up with AI, the podcast where human powered meets AI assisted. Join your hosts, Chris Carillon, Niko Lofakas, and George b Thomas as we dive deep into the world of artificial intelligence. From the latest AI news to cutting edge tools and skill sets, we are here to help business owners, marketers, and everyday individuals unlock their full potential with the power of AI. Let's get started.

Good morning. Happy Tuesday, November 12, 2024. Feels like so long since we met 4 days ago. Time for another episode of wake up with AI. How you doing, George b Thomas and Niko Lafakas?

Man. I'm, doing good. Loved the long weekend, by the way. Tried to relax a little bit, chill out a little bit, not really think about too much around AI. Although, that was almost impossible.

Oh, that's

a good one, George.

That's

Yeah. Did you like that joke? I'm here all week. What's fun though is just when I got back to our Slack this morning, I was like, oh, there's a new video. Let me go ahead and let me go ahead and throw this in there before the show.

5 hours and 15 minutes of a video. I said, good god. So I started playing it at 1.25. It's very exhilarating. I hope that we're gonna talk about or mention or maybe put it in the show notes, but I'm doing okay.

But, man, is it hard to keep up? Even when you're waking up every morning with AI, it's still hard to keep up. I'd

echo the same statement entirely. I I will admit that was foul play on my part. I saw that video get posted, which I blame Lex Friedman really for posting that on a Monday, but I understand more traffic that way. But, yeah, I was watching it yesterday and then thought to myself like, oh, this will be cute. I'll drop this to the channel at the border, and the guys will be like, wow.

Really? Like, what are we supposed to do? Like, oh, how could we get this in time? So

This is this is why I'm gonna today work with my AI to try to build a time machine. Because if AI can help me build a time machine, I can actually get these videos, go back in time, watch them or listen to them, then come back to where I'm supposed to be. And it'll be like, it'll be a beautiful world when my AI helps me build a time machine.

Yeah. If we ever build 1, it will be because AI helped for sure.

That's definitely going to be the case. Hands down.

So what are we talking about today, Nico?

So today, we'll definitely I'll I'll touch on that that video. So what George was referring to is a video from Lex Friedman from Alexis Podcast, and he's interviewing Dario Amede, who is the CEO of Anthropic. Him and his sister, run the company together. So in this interview, Dario is talking about obviously all things AI. And the reason that these interviews are so impactful and so important, in other words, when you can find a YouTube video that is featuring either Dario or it's featuring Sam Altman or it's featuring Jan Lecun or, you know, any of the prominent people from the AI sphere, Demosysavus most especially, you will definitely or Mo Gadda I mean there are so many I'm sorry I could just keep going on you're getting the real story like these are the people I listen to these are the these are the videos and the talks that I watch because these are the guys who are at the forefront they're building the models I don't listen to the 2nd rate people I don't listen to the second hand people I don't listen to the financialists I don't listen to any of that because those guys are second third like you know in most cases third second would be like people who work at the company to me and then third is people outside of the company So I can't listen to that because it's it's a lot of fuzz.

It's a lot of traffic. The you know, it's like playing telephone. So by the time the media touches it, it's a whole different thing. And so listening to Dario this morning is just obviously very eye opening, very enlightening. It seems like what he had said earlier this year was they were gonna do a 3 5 opus release this year.

They were gonna do a new now that Sonic had released, they were like, yeah, we're gonna do 3 5 opus. Probably by the end of the year, we'll have a new one. It doesn't sound like that now. It sounds like it might be early next year instead of late this year, and I can fully understand why. Again, if you've been working with SONET, if you've been using SONET, from Dario's own mouth, it's the best coding platform in the world.

I mean it's it's the underlying model for Cursor, like there's there there are a lot of tools for coding that are built on top of Anthropic's model, and there's good reason. You know, I've been sharing internally at the fact that just with a little bit of a very small amount of training and I have no qualms admitting that that one model might be better than my custom GPT. My custom GPT is old. It's it's old news. Right?

It's a grandpa by comparison to what's going on right now. So just with a little bit of work, if you build up a quad project and you give it some custom coded action snippets that you've done or that you're working on, you just give it some examples of that, and then ask it to help you build a new custom coded action, especially if it, you know, makes one correctly already, it will start one shotting them.

Guys, I gotta I gotta share a story because you're talking about this. Yesterday, I had one meeting because I forgot to block it off my calendar. It was a training, and I was training somebody on HubSpot, and they said they wanted to learn about workflows. When we got on the training, it was actually that they had a custom coded workflow that wasn't working properly. It was a custom coded workflow that was supposed to send an SMS, but they wanted an image to also show up in the SMS.

I said, well, this is probably above me. You might need a developer, but hang on a second. And I took the code that they had in there, and I went over to an AI model to, not be named at this point in time. And I said, hey. I've got this code, but I wanna send a text message with an image.

Here's the image URL in the HubSpot file manager. Can you rewrite the code? By the way, they thought this was impossible. It wasn't gonna be able to be done, and they were so frustrated with HubSpot. So we get the code, we put the code, and we test it, and it works.

And they said, but hang on. There's this word where it says image added or something like this phrase. And I said, well, hang on. Let me ask my professional coder. Is there a way for us to remove these two words?

And the the coder came back and said, yes. You can do this and spit out the new code. I put the new code, and I tested it. And I they they got the text, and they're like, this is beautiful. This is perfect.

By the way, they loved me. They thought I was awesome, and I was like, oh, I love my assistant. My buddy is awesome. Like so the power that you're talking about, Nico, like, I just wanna give a real world example of, I don't know that stuff. I'm not a coder, but it walked me through and gave me the ability to create an experience that people went from frustrated to joyous in, like, 7 minutes.

Think about it, guys. You're you're working in and and this is kinda how I look at it. I'm sure you you guys look at it the same way. We're working in HubSpot all the time. The limitations with workflows is just that, it's limitations.

The reasons for me to get people to upgrade to Opshub, really kinda minimal. Right? Because the cleanup tools, even through workflows, minimal, and the cleanup tools that you get from data quality, Okay. The the reports are more leverageable than the actual data quality tools in some cases, but it's getting better. It it is.

It's getting there. The filters now are are making it a little bit better. But still, it's a matter of, like, if I could custom code the action to do the cleanup for me, this would be so much easier and so much faster. So think about it from from rev ops, from from your HubSpot, from an admin perspective. You have just increased your toolset infinitely.

As long as the API endpoint exists, you have increased your capabilities and your toolset infinitely to a very sick degree. I have already built custom coded stuff for a client that doesn't exist on any other platform, and I've taken it to a much much further degree where quite literally one action in the CRM will trigger a whole lot of stuff going on between records getting updated, others records getting removed if quantities are changed, others other records getting created if quantities are changed, and the created, records incrementing accordingly. It's very, very complex. Or, you know, if I had to ask a developer to do first of all, our internal development team is is busy. They're busy doing stuff.

So that becomes a list on their items of to do. That pushes the project time back, or I use my assistant and, you know, I I do it as fast as I can. So to your point, it's it's just an extension. Like, the capabilities of anybody any admin right now, If you're not getting into custom coded actions with this stuff, I don't know what you're doing. You're wasting your time.

If you're doing like, honestly, if you're doing workflows and you have Opshub Pro and you don't have a quad subscription, you are wasting your time. There is a there is so much you can do in the portal right now, way more than you ever thought possible. Right? And so you turn you end up turning into I don't wanna say you turn into a junior developer. I don't wanna take that away just yet, but it's something like that where you now add this layer of coding development skill to what you're able to do without necessarily having to go out and grab that extra resource.

If there's any spot to wake up to, it's here. Like, whether you're an admin, whether you're somebody developing product for people using HubSpot or any like, we're talking about HubSpot, but it's really any system. Right now, like, coding, all the different types of coding languages are just that, those languages. And when you have access to something that speaks or uses that language very well, like an a s I like an AI assistant, and we talk about when there's perfect information available for everybody and you're developing products or you're developing managed services that you think you're gonna make money on because people are gonna have to come to you to do the coding work. They're just not.

They don't have to. Right? And these tools are available with the knowledge. Even when new things become available, like one of the updates I mentioned today is there's a private beta available for creation of invoices. A lot of people have been waiting for that.

Right? It's been read only. Now create is there. If you've got this, like maybe you weren't even considering Like, I'm not even gonna touch invoices until that thing becomes available, so you haven't been building up any kind of skills related to how the API is gonna work or, you know, how are my custom coded workflow is gonna work. Now, because it's all in the same language, you're just like, hey, AI, invoices are I can create invoices now.

Let's let's put a custom code workflow together, and this is what I wanna do. And it'll the difference between 0 shot, 1 shot, 2 shot, right, is is getting is getting is getting smaller and smaller where, you know, we also heard, a use case yesterday, you know, from, you know, friend of the pod, Alexia, in that she was having some challenges. And, you know, she reached out to Nico, and Nico's like, yep. Just a little bit of this, a little bit of that. Right?

Using these multiple tools. And, of course, she's got a day job. She's not just digging into workflows and and AI every day. So over this over the course of a week, she was able to go back and forth enough to say, yep. Got it.

This thing that I thought I wasn't able to do, I can now do it. And I did not have to pay 1,000 of dollars and spend even more time working with somebody else so that they understand my business and can get in there and freak everybody else out in the company in the process. Right? Like, who's in here? Who why are they in here?

What are they doing? Right? Now all that stuff goes away.

See, and and a little garlic, a little paprika, a little AI spice, and things are gonna change. What's funny is this is a great segue for that's right, ladies and gentlemen. It's time for another AI skills that pay the bills because today, I wanna dig into a skill that's essential, and we've been talking about it. We've been dancing around it today. AI as it continues to evolve our work lives.

So let's talk about flexibility in AI role evolution. Or in simpler terms, being ready to shift gears as AI reshapes what we do and what we can do. Here's the reality. AI is changing jobs. It's taking over repetitive tasks, automating data work, even personalizing customer experiences, but guess what?

This doesn't mean we're becoming less valuable. We've just shared a couple use cases where we have become more valuable because of the change. In fact, it's the opposite. Our roles are evolving to make space for deeper, more strategic, and more creative work. But to stay relevant, we need to be flexible, ready to adapt to new responsibilities as AI shifts the landscape of the work that we're used to doing.

Think about it like this. Imagine you've been handling data entry for years, and suddenly, there's an AI tool that can do it in seconds. Now instead of panicking, flexibility means you're saying, alright. What can I do now that I'm not tied down by these repetitive tasks that I have had to do for years? You start focusing on analysis, creative strategy, or connecting with customers on a more, dare I say, human and personal level.

That's the evolution in action. Flexibility in AI role evaluation is about seeing these changes not as job loss, but as job transformations. It's about being willing to learn, grow, and take on new challenges that only you, the humans, can handle. So instead of seeing AI as competition, see it as an opportunity to level up, well, you and your humanness. And here are 3 takeaways on flexibility in AI role evolution.

Embrace new skills, ladies and gentlemen. I wanna say that, like, 17 times, but we don't have time. Be open to learning new things like analysis, strategy, or relationship building. Focus on human centric tasks. AI can handle the routine, so lean into areas that require creativity and human touch and see AI as an opportunity, not a threat.

Let AI handle the repetitive stuff and look for ways that frees you to grow in other areas. Remember, flexibility isn't just about going with the flow. It's about actively steering your career towards new and exciting possibilities as AI opens up the doors. Keep evolving. Keep learning.

And that's a wrap for today's AI skills that pay the bills.

The flexibility aspect of it is not that it directly relates to, like, workplace flexibility, but it did kinda bring me or at least rebuy me of the story it came across 2 days ago where it's flexibility in thinking. A robot did a painting, and the painting was of Alan Turing, and it was a very unique painting. I'll say that much. Like, the way that it looks is extremely unique. The estimates initially were that it was gonna sell maybe for around you know, this is painting on canvas.

Right? It's a robot that painted on a canvas. Estimates were it was gonna sell for around, like, maybe 200, 200 grand tops. Right? 1,100,000.

Which is crazy, by the way.

And a lot of that might be I I can assume that a lot of that is I own the first, you know, painted by robot painting. So that I can understand as as being part of the price tag. But just the fact that they were saying that it was expected to sell between a 120 to a 180,000, like, that to me is like, even that is interesting.

Interesting. That's some people's, like, annual salary for 4 years. What do you mean interesting? Like Yeah. Right.

Wake up

people. Let's face it. Especially when it comes to art, beauty is in the eye of the beholder, and what moves you is what moves you. So people are willing to pay what they think that it's worth. Right?

You know, I don't I don't know how that works out. Like, realistically, I don't know if a lot of that is based on, you know, the assumed skill level that it took, like, the type of training that it may have taken. I don't know if it's based on the image itself. I I don't know. Or it's just based on emotional reaction.

Some bebop boop shit, and all of a sudden, there was some painting on a canvas. Like, the fact that it went for that price where it was estimated for that price, I believe that you were hitting nail on the head at the beginning. It's like the first, which by the way, first mover advantage has always been a real world scenario. Like, why am I who I am today? Because I started a HubSpot in 2012, you know, using HubSpot.

I don't work for HubSpot, but 2012, early adopter. First, why do people like, why do we talk about podcasting, early adopter? Video editing, early adopter? AI, early adopter? Like so it's like the early it's the first.

So therefore like, do you remember the the, like, oh, anyway, never mind. This topic right here frustrates the crap out of me.

But what is And well, this is where, like, the estimation. First of all, the estimated dollar amount. And it takes me back to last week where we're going through that port, and all these experts are proving to be terrible at estimating AI progression and how fast it's gonna progress. And first of all, humans, they're not gonna do a good job of this. One, because this is brand new for everybody.

They want control. They wanna protect themselves. So I guarantee some of that low value was like, please, Lord. Let's not let this be more valuable than the works that we do ourselves. Instead of, like, all the things that the other variables, like, oh, this is the first one.

Oh, there's people that are gonna appreciate AI art probably more so than human art. Maybe there's new people that will appreciate it just because it's AI, and they don't give a shit about all of the variables that you would usually care about to sell, like, the most, you know, elegant piece of artwork. There's all these elements where humans have an agenda. Like, subconsciously or consciously, we do. So that's going to come through in any kind of, you know, estimation or prediction or forecast or whatever.

That's where, like, I love getting information from the longer form content and conversations. Like what you shared, I've always loved that from founders, especially, like, you know, Dharmesh and Brian where you'd listen to these hour long interviews about how things have gone. You just start to get these nuggets that you're not gonna get anywhere else. And when you can help yourself understand the thoughts, the thoughts and intentions behind the scenes. Like, that starts to unlock 1 more trust in some situations, less trust in others.

Like, okay. This is a very you know, they have a very intentional way that they're going about this, and it's, you know, sometimes profits driven depending on the leaders of these companies. But that's where you know, listen it to Sam and Dario. It's like, I don't get any of that from them. They are just like, let's get this powerful, valuable thing as ready as possible for as many people as possible, as soon as possible, as safe as possible, depending on which one you're listening to.

Yeah.

Yeah. That depends. That last part depends, bro,

who you're talking to. At least, like, when it comes to Anthropic, I very honestly don't worry. And that's because they not only use, like, the US AI Council, but they also use the UK AI Council. And I'm sure once the EU put puts one together, they'll go through that as well. In other words, they are, like, happy to give their model up for people to test to try to find if there's something wrong with it.

Right? Because they they know better to find it early than too late. Right? And, you know, we may not see an Opus 3.5 release this year, which should speak volumes. If you're a Claude user and you use 3 and you use Sonnet versus Opus in that level, do you understand why they might be hesitant about releasing a 3.5 version of Opus?

It's just it's that much more realistically. It's that much more reasoning. It's that much more capability that leads to, you know, which Sam has talked about before. So, like, a lot of people, you know, Salesforce talking about agents, HubSpot talking about agents. There's a lot of, like, no code agent builders out there.

I just wanna distinguish the difference between what these guys are calling agents and what the forefront calls agents. The ones that you're seeing right now that exist, those are not agents. Those are just, like, advanced workflows is what you could consider those to be. You're building the chain of thought. That's really all you're doing.

An agent, realistically, even Altman has said the same thing, can't really exist without AGI. It can't exist without general intelligence because it must understand how to do the steps in between so that you are not giving it. So here's the difference. Currently, you're building a chain of thought through workflow actions. When real agents are released, you will give it the end aspect just the goal.

And I I think I've shown this to George or or or Chris. You'll give it the end aspect. It will go do all of the steps in between. You won't have to ask it to do any additional stuff. You'll say, you know, hey, I know everybody uses the travel example which Sam also hates, but you'll say, you know, hey, I wanna do some, you know, deep research on a particular subject or something like that.

It'll go, and for days or however long it takes, we'll go do that task and then come back. And it'll even you know, it might come back to you if it, you know, stops and has a very complicated question, or maybe there's a preferential question that it needs your answer on a preference. It's a monumental difference, and you'll know it because it's coming next year. All these guys, like, they they all say the same thing. They're all so whoever was like, oh, AGI, even myself, right, Rakers wheel AGI by 28, nope, next year by 26.

So the estimates on Superintelligent were 35 to 40, so I can easily say 30, if not sooner. If we have AGI next year, it'll be sooner. It just will be. It won't be publicly available sooner, but the capability will be there sooner because every iteration is faster than the one that preceded it.

Honestly, I can't wait because, again, a real world scenario. I had a hour long meeting with a client. He did some brainstorming after our call. He sent me a document with, like, I don't know, 3 things and 7 bullet points. I then had a meeting with him the next day, and I said, hey.

I wanna show you something. What do you think of these 5 programs that you could provide? And this is around, like, leadership and coaching and, like, teams and individuals. And he was like, love it. Love it.

Love it. Oh, I love all these. These are amazing. Sweet. And then the next thing I'm gonna do, I'm gonna actually flush them out to, like, what the 90 days would look like, what the 12 months would look like.

Ladies and gentlemen, I tell you, I generated these. I put them in a word docs. I sent them over to him. Here's what the emails look like. Approved.

Approved. Approved. Now he knew the caveat was I'm sending all of this over as drafts, and we can adjust the times and the places and virtual or whatever. But, like, every single thing was approved, and literally, we saved him maybe a year of trying to develop 7 programs for individuals and teams and modern leadership because Claude was freaking

Wow.

Dope. Like and so, like, that's why I say I can't wait. Right? I can't wait till it gets to the next level, and we can get even better and do even better and do even more. Because I know that the amount of humans that human will impact because of the rate in which we could leverage Claude to actually give him the foundation of where he was headed based on what he already knows from the rest of his life.

So it's like plug this into this and roll.

And we can't wait, and we aren't waiting. That's why we're here. Like, there's a learning curve here, especially in the cases where it's like, check this draft, review this, figuring out when and how to do that effectively so that you don't, like, over bias stuff, so that you don't approve stuff too early, but you don't wait too long either. There's so much of this that leads into, you know, the skill that we talked about of being flexible, like agility, speed. We don't know when this is gonna hit.

But those who are ready for it, when it hits, will have a, like, just the biggest advantage. Like, and those advantages are only getting bigger and bigger. So if you're not one of those that is just honing your AI skills every day, trying out tools, experimenting, understanding how to interact with it with all of the different possible work that you're doing, it is time to wake up with AI. Have a great Tuesday, everybody.

That's a wrap for this episode of wake up with AI. We hope that you feel a little more inspired, a little more informed, and a whole lot more excited about how AI can augment your life and business. Always remember that this journey is just the beginning, and that we are right here with you every step of the way. If you love today's episode, don't forget to subscribe, share, and leave a review. You can also connect with us on social media to stay updated with all things AI.

Until next time. Stay curious, stay empowered, and wake up with AI.