AI,AI, AI - The Last Mile

Yeah. Good morning, LinkedIn friends, HubSpot Nation. Maybe you're on YouTube watching another episode of let's Build AI Skills with Nico, George and Chris. Another week of. Of fun and excitement in the AI world. Gentlemen, how you doing today?

Is that what it's called?

It's a similar energy to anxiety and other stuff, too.

Oh, my God.

Energy is energy, right? No, I'm doing good. I'm doing good.

Yeah. So interesting story for me last weekend, and Nico and I were just talking about, like, somehow we can find ourselves with little time, even though we have all of this AI stuff going on, because we're here trying to build bridges, right? We're not just trying to get the old stuff done. A challenge I've always had is, like, project management for myself and just keeping organized. So finally I was like, I hear this MCP thing. It's pretty cool. I see use cases where people are plugging it into, you know, their Google workspace, and it's handling calendar and tasks and. And all the stuff, right? And I had already had an executive assistant, you know, Claude project that's, like, set up and ready to do that stuff for me, but it's never been connected to the actual schedule and the actual email inbox and that kind of stuff. So I was like, all right, wake up on Saturday. And like, let's do it. Like, I'm just gonna start. I think I was in my bed still, like, pull up Claude on the phone. Like, yeah, it's just like, how are we gonna do this, buddy? And he's like, finding, you know, MCP servers. He's. And so then I get my laptop out, start downloading, start following all the steps, get 95%, and then we can't connect to Google, we can't see Google. Everything else is working. He's trying his best. I have no dev experience, so if Claude can't help me in those moments, I've got nothing. Right. And I'm hearing, you know, more stories like that. But I'm just curious how you guys are approaching that, especially when helping others, right? Because it comes up with vibe coding. It happened to me. I think I showed it with Manus last time. Like, all right, we get to the website publishing, and then, you know, something's missing, right? So how are you guys. Is that resonating with kind of the tools you guys are using, the learning curve you're having and seeing around you a bit?

Well, actually a lot. And for the Manus aspect of it, like that. So I can understand the limitations there with operator Right. Because you're working with a, within a virtual environment. So technically you, you can't really launch anything there that you should be able to view outside of that. But it is possible, right? Assuming that you're, you're able to do all pure login and like non Google login, you could like log into your services and then have Manus do that. Right? Like operate your, your various services and probably launch your page for you. But it's stuff like that. So like I was, I was talking to somebody on the boards yesterday and having this exact discussion where they were saying the exact same thing that yeah, I can't really get as far as I want to. It seems like all I can develop is like a playful front end. But you know, after that it kind of just goes south and I had the same experience myself a couple weeks ago and was getting super, super frustrated. Have tried, I've tried just about every vibe coder out there. If it exists, I've probably tried it. If, if you, if I haven't tried it, send it to me and I'll check it out. So far none of them hold a candle to Replit. That's just, it's, it really is on a whole other level it does matter how you work with them. So I have adopted this new framework of, you know, having an AI team and what my AI team looks like. And so it's this combination of using Claude to start the development process and Claude becomes, you know, like I'll use GPT or Gemini as my researcher. That comes into Claude, who is my front end developer and then that goes over to Replit, who is my, you know, developer overall full stack dev. Right. And then it's like a bit of a back and forth. There's also a matter of like knowing how to use the tools in terms of some. So like Lovable doesn't have the option for like multiple conversations. It's just one long conversation. Now that really sucks because especially in Lovable, you're not building. I wouldn't use that to attempt to do anything that remotely needs something related to a backend source or resource or anything like that. You're overstretching it. Simple web application or simple mobile app? Yeah, Lovable is really great for that. Concepting, prototyping, really good for that. But what ends up happening is when you have to make all those changes and stuff, you start opening up that context window and after a while again it just starts going south because it forgets what you're even working on in the first place. I had it randomly changing the design just because it couldn't seem like it couldn't remember what we were working on. So it would make an update, but then change the design completely. Whereas in replit, what you can do is you have agent conversations and so you can start a new conversation. Say like every time you make a large completion to where there's like a rollback setting, you can now just start a new conversation. And that's like, okay, now we're moving on to this next part of it, but there's this history of what we have worked on in the past. Windrunner has something like that as well. But I. I still prefer Replic.

Yeah. And before we get get to you, George, I just want to highlight. And most people know I only promote stuff and us like when I know it's going to be valuable, like for sure. And let me tell you, I've had several people unsolicitedly say to me, did I tell you I met with Nico? Like, I took him up, I paid him for an hour and wow, blew my mind. And this. So this topic that Nico just brought up of building your AI team, right, like, only can come from like trying all these tools and understanding how they all work together. But that is a big part of the learning curve that, you know, even me, like, I'm trying to just ram everything through Claude right now as much as I can, even though I hear great things about Gemini and that's right in the Google workspace. So why the hell am I not trying to use that? So maybe from that perspective, George, like this concept of maybe your learning curve as you've tried to push each tool and you run into roadblocks and kind of your mindset related to that.

Yeah, it's interesting because I'm very much less the coder side. I'm not, I'm not vibing much except for life itself, to be honest with you. Like, and. But when I think about the creative side of this, there is for sure where I'll use AI to a point and then I'll end up in Photoshop, or I'll use AI to a point and I'll end up in Canva or something like that. Or there's a point where I'll use A.I. and then I'll use A.I. and then I'll use A.I. right? And so like, it's funny because some people get hung up on this thing of like, it's got to be all in the same conversation. It doesn't. It doesn't. And so like last night, Chris, it's funny you're talking about using AI in bed. Yesterday I had this spell where I just didn't feel well. And so I literally went and like laid down early, fell asleep, and then couldn't fall asleep at night when I usually do. And so I was up messing around with AI, being stupid and creative. And I shared some things with you, but that was like three or four conversations deep to get to the things that I was showing you, because what Conversation one was getting the idea to a point where I could then take it to conversation two to layer another layer on it to get into Conversation three to take it in a totally different direction, but based on the base that we had been building and. And so like breaking all norms. Like my. My brain is such. In a place where I don't believe anything historical other than the biblical truth of like, fundamental life core principles to be true anymore, because. What do you mean I can't write a book in two hours. Yes, I can. What do you mean I can't use AI and actually add value? Yes, I can. Like, at the end of the day, I was on a meeting yesterday, guys, and I explained one little thing and you would have thought it blew their mind. And I was talking to them about this concept of building blocks. When you have the building block of values and you have the building block of mindsets and you have the building blocks of belief. And also you take that and you throw in Tron legacy and you throw in this and you tell it to write an article. And I shared the article with them and they're like, oh, my God. But you gotta. So when I think of the last mile, it's right here. It's opening your mind to what is now possible based on the world and tech that we now live in and with.

Yeah.

So break what you think is real, break what you think is normal. And just. I swear to God, this is the biggest superpower at this point. Is it possible?

Yes. Oh, my gosh. And like, oh. On the other side of. Of tech, like, and this is so much what. What Value first is trying to accomplish is like everything we knew about B2B and SAS. Throw it out the damn window. Because I'm sitting on a call yesterday and thankfully I get to have this masterclass right now of just watching. You know, we need to do an integration for an ERP. There's iPass platform in the middle and there's different partners for each of the pieces and parts. And there is a native sync issue between HubSpot and NetSuite and it's causing a company ID that is unformatted numbers to show up as formatted with commas in NetSuite and 1500 installs. So it's not like nobody's using this. This sync. And, you know, I had to be the guy on the call saying, okay, so we. We developed another workflow to transform the number work around the comma problem and created another field in netsuite to put stuff in. I'm like, has anybody talked to HubSpot or NetSuite yet? Like, what are we doing here? And what we are doing there is the, you know, the. The other third party. They're used to this, right? They've got a job to do. They've got a contract to fulfill. They see crap like this happening all the time in the systems, and they're ready to work around. They just default to the workarounds instead of trying to solve these problems. And that's why, like, that's at the heart of, I think, what George just said about so much of what we know. And, you know, the client who was also on the call wants so badly to seed, like, to the best practice judgment of the person he's paying to do the work. And when the best practice judgment is workarounds, because that's how we get our job done. We are not in a good place, folks. The other thing that. That came up while you're saying that, George, like, because, like, literally, I'm so excited for a future where teams can work together with this AI conversational layer. I haven't really seen it. I've tried to accomplish something similar with let's build, you know, let's build Claude instance.

Yeah.

But we're not quite there yet. And so George sends these pictures over. It gives me an idea to say, all right, I'm gonna take the profoundly logo and I'm gonna take, you know, the. The Hulk, the smart Hulk drawing of the caricature of me that George made a couple years ago. And I'm gonna ask to put them together. And then it did that, and it gave me all, like, elf ears or troll ears, because apparently if you're bald and green, you gotta have troll ears. And then I'm like, okay, GPT. Like, how about my pictures? Here's the reference. And then it. And then it gets a lot better, but changes the color, so it's way ahead. And then George comes back with, clearly. I'm pretty sure that sparked an idea to have him share it.

I know you can see the screen down there that I'm trying To share. There we go. There we go.

All right.

I had. You have to have the image to see, like, but. But here's the thing, Chris. I didn't care about the ears. I saw this and I thought this is the coolest ish ever. Which, by the way, I had shared these. Like, I was working on some. Creating HubSpot super admins worldwide, right? And so Chris. Chris sent me this. But then you got to this, bro, and I was like, oh, oh, let's go. Like.

It was fun. And then did that. I'm making an assumption here that that sparked you to infuse your own likeness into the images you were in the middle of, right?

Yep, yep.

And that, man. Like, this is the kind of stuff that, like, usually only happens during, like, verbal. This is why those brainstorming, like, sessions where you got. Where the team just gets in zone, like, ideas going back and forth, but we're not creating in those moments. Generally. Yeah, you gotta do all this admin work to try and capture the idea.

So, man. So I have to. So I can't. I don't have a way to share this, but I look forward to the day and maybe it's happening. I. Maybe it's happening, but, like, I. I miss working in an office where it could have been like, three or four of us huddled over a. A screen. Well, what if? Yeah, but what if? What if, what if, what if, what if, what if, what if, what if? And. And I. I literally was having this moment last night, not at a desk, but again from bed, because I was texting. One of my daughters is my social media manager. One of my daughters does a lot of stuff around content, and I was sending them, like, icons with, like, solids and gradients and then this and that and superhero and not. And they're like, yeah, but what if we did the blue and the green? Or. Or what if it was the purple and the orange and, like, can you try this? And we're rapidly prototyping, like, what's the new set of icons or graphics that we want to use when we redo Sidekick Strategies January 1st of next year.

Yeah, it's the same. Like, I don't know if you've tried this yet. It's pretty. Pretty sick, actually. Yeah, I can demo it, and it's actually pretty fast too. So I was trying to come up with, like, just designs for applications in general, and. And I was like, well, you know what would be cool is, like, coming up with the UI and everything, but I'm not the greatest UI designer either, so. It's like pretty difficult to come up with that stuff sometimes. So what is easy though is even, even though I don't have a membership, I think this will work perfectly.

We're hoping. Yeah. I think what we're talking about here, like the three of us have different last miles and like at the end of the day we didn't get to a point where it's like AI plus human, the whole miles, all the miles, like back and forth. Like, you know, some of us are going to be able to get, you know, like George can go into Photoshop. I can't. I don't have any Photoshop skills, you know, to finish that job. Nico, on the dev side, right. I was on a show yesterday though that was talking about, you know, building out virtual environments, you know, for teams and organizations to work in to get used to, you know, change and start to understand what it's going to be like to have wearables on and AI involved at this, like, you know, integrated collaborative layer. So like one is opening the mind. Like you've got to get rid of. That's the thing. It's hard when you've got to deal with partners and other people. You're dependent on, you know, third parties to manage systems. Like everybody's in this place where we're working off of old infrastructure and it can be hard, I, I imagine. Well, no, it is hard. I have to deal with it myself, you know, day to day. So maybe after we check this out. Some advice, George. How do you keep an open mind but also stay grounded in that the people that you're likely to run into to get stuff done on a day to day are likely not in that place. You're muted.

He's like, no, I, I'm sitting here thinking, because I just want you to do one little thing for me and rephrase the question that you asked. It's not that hard, right? How do you get people to have an open mind or how do you keep an open mind? That's what you asked me. How do you keep an open mind, bro? Do we have four hours? Like so, so like it's, it's a whole mindset.

You.

There's closed mindset and there's open mindset, right? And, and it's a choice, but it's also a battle. Like you have to put the micro pieces and so without. I'm not trying to go anywhere with what I'm about to say, but this is literally the superhuman framework and the cornerstones and the pillars and the things that I'VE broken down in my life to pay attention to, to. To wake up and have programmed my mind in the morning to wake up with an open mindset. Right? It's, it's zero expectations, it's zero judgment. It's loving everybody in all things. It's like, it's like walking out of the bedroom with a, with a, an emotion of embrace versus, like. Dude, I could tell I don't have four hours right now.

No, you make a good point. I think that's like, where you reminded me of what actually happened on this call yesterday. Right? Because when I say that out loud, I'm like, what are we doing here, folks? Like, we're literally developing more complexity for a problem we haven't even tried to solve, like, in a different way. And, and the third party on the call, he was like, yeah, I totally agree. We should not be settling for what's happening right now. Right? And it was just like, I give him the space to be like, dude, stop doing what you're doing. Like, it's not okay. Right?

So it's, it's interesting we had very similar conversations with different humans yesterday, because these were some exact words that came out of my mouth yesterday. Listen, I get it. Sometimes we humans like to over engineer stuff. Sometimes we make open mind more difficult than it needs to be. Sometimes we make value first more difficult than it needs to be. Sometimes we make waking up with AI more difficult than it needs to be. Stop it.

Yep. What's funny about that is I've caught like, even the models doing that lately, and it's. I'm not sure if that's something that is becoming more widespread or, or not or if anybody else is experiencing the same thing.

But.

Lately, especially like with the reasoning models, I'm just, I'm getting like, answers back where I'm having to actually tell the model. Like, hey, I, I think you're overshooting it. Like, you're, you're kind of going beyond what it is that, like, we're actually looking for here. Like, you, you really should just kind of try to like, even what happened here. Right? Like, I don't know how much extra code it wrote, but it was clearly too much. Right?

Yeah. Like, Bill shared a story with me where he's trying to set up like a scoping spec project and it's quoting like, know 120 extra hours based on the stuff it was given. Like, that's where it's like any kind of project management timelines. Like, if I don't tell it to like, get rid of the human training that it's gotten, right? It gives me the longest stuff and it's like, I'm like, no, Claude, we're in. This AI thing is happening right now. And I know you can go a lot faster than this. So it's, it turns into like days from months, right? Because so much of it's just like paperwork and alignment and having meetings and all the, so much of it's just like paperwork and alignment and having meetings and all this stuff. And it's just like, we gotta find a way to cut through that stuff.

It's gonna, it's gonna take some time. And I know that people are already struggling with the idea of, like, face value. So if you're wondering, like, what it is that you're looking at or, like, why you're looking at it, basically what I did was went to Disney's website for Tron, took the link to the website, gave it to Claude and asked it to build out a web application. So it did. And then it clearly doesn't look anything like the website. So I gave it the pictures reference again, told it to match it up. So I tried a little too hard, tried to write something probably super complex. And so I just told it to simplify it. And here we are. So three prompts and seven minutes or something like that, and you've got a prototype that looks like whatever it is, reference, like image, reference wise. Right? So, you know, to George's point, it's like you can, you can prototype stuff so fast now, you can ideate so fast that, you know, for us it's, it's a different story. Like, for us, we talk about not having enough time because we, because we are spilling all the ideas out of our heads and trying to bridge all these gaps and trying to make all these things happen. And it would be like, I would love to see what that looks like when it's, you know, widespread, when everybody is doing that and everybody is sharing their ideas and everybody is, is working in this sort of fashion. And I, I, that's what I see as being that great positive driving force with AI, is people finally getting involved in being able to expand their, their creative thinking, their critical thinking and being able to contribute towards things. Right? Because, like, if you're a producer, there's nothing wrong with it, right? Like, we've been in this production level for such a long time, but to produce, it's not, it's not on an individual basis. Not always. In most cases, it's not on an individual basis, which is why most people feel unfulfilled at their jobs. Because you go to work, you know that there's a pile of work. You know there's going to be a pile of work tomorrow. And you know that when you do the piles of work, you personally don't see the benefit of it. Rarely do you see the benefit of it. Even if your rev ops working on somebody's portal, you're not doing a check in six months. If you're really good, you are, but you're not. Most, most of you guys are not doing check ins six to eight months later. Asking them how the revenue is, asking them how all the, the mods went, asking them how, how it is. Right. That would be trying to attain some of that, you know, satisfaction in the work that you did. But when it comes to contribution, you get something that all of you guys desperately want and seek so much. Which is why this stuff is so, I don't want to call it addicting, but it, it kind of is. But it's in such a positive way because you're actually addicted to something productive, educational, like it's hard to walk away from it not learning something. Yeah. So yeah, I mean like contribution is where things are going. Right. And watching the rest of society be able to contribute what it is that's in their heads towards what it is that we're doing. Yeah, that's, that's how we move forward. Right. Is by sharing ideas.

Here's the thing. When it comes down to the future and creativity and communication and community and collaboration and AI and what we're. Where we're headed, if we do it right is we're headed in a world where we're going to buy back enough time that we actually can care that you can do the check ins in three months, you can do the check ins in six months because you have the time back. You're off the dang treadmill. You're not the hamster on a wheel because. Because you're living in a different world. Because you're embracing the power that's at your fingertips. Or you're not.

But you're about to be. So please, for the love of all things holy, find those opportunities and let's grow. Have a great weekend, everybody.

See you guys.