What does AI Vibe Coding have to do with your CRM?
Chris, George B. Thomas, and Nico argue that AI agents should automate work outcomes, not recreate SaaS tools like CRMs. Chris built 21 courses with AI, generated audio, and assembled a learning platform, but notes that replicating HubSpot's 14-year ecosystem would be impractical. The real shift is toward AI agents as personal digital replicas handling data and APIs autonomously.
Chapters
Sam. Good morning, LinkedIn friends. Value First Nation. Welcome to another episode of Value First AI Daily, your collaborative intelligence report. It is Tuesday, February 10, 2026. Happy Tuesday, fellas.
Happy Tuesday.
Happy Tuesday. So you got some new digs over there?
Yeah, getting. Trying to get settled in. I had a fun weekend of moving.
And.
I had AI with me along the way, helping me unpack mentally and physically.
Did you leverage Gemini to help you rearrange furniture? Instead of.
Gemini is helping build an inventory of both mental and physical inventory related to all this stuff being moved. And I think in my. In his first response, he was like, yeah, you know, moving can be one of the most, you know, like, emotional things with, like, you finding things and reflection and all this stuff. I was like, yeah. And since I was moving by myself, Isabelle's in. In. In Columbia, you know, it was like I wasn't moving by myself. And I think it's things like that. Like, I hadn't moved. I'd moved. Like, we moved like six times in four years before we had the house. You know, I have ADHD and I don't mind the. And I kind of like the change of scenery, but I hadn't done it. Last time we moved was August of 2022.
Wow.
Right before GPT came out and this whole AI thing started. So it was the first time I moved. And even there, it's like there's a different way that I could allow AI in to help with various things. Um, but I need to dive in because we got started a little late. Uh, yeah. So I haven't been on LinkedIn as much over the past few days, but every, every time that I have logged in, I have seen a post about why it's not a great idea to vibe code a CRM and all the reasons why. You know, HubSpot's going to be fine. Software is going to be fine. Like, all of this in response to the SaaS apocalypse stuff from, from last week and oh, a message to the first regiment of the. Of the CRM army that is. Is taking. Like you're missing the point of what's happening. And like, I'm not. And many others are not vibe coding CRMs. We're not vibrating project management systems. We're not vibe coding cmss. We're vibe coding the work and the results. And we are using agents to bypass software altogether. Right. And like, so if that's where you're at in terms of the message, like, it's just. I get where it's coming from, but it's. It Feels like a, like a misplaced energy, I'll say. Because I don't think anybody wants to buy code most any software system. Like especially once you get access to the cloud code and the AI native stack. Like it's like, yeah, let me just ask this friendly AI agent to help me achieve this output, this outcome, this goal and then maybe it tells me to go get, you know, a new software account. More and more likely that it does not. So that's, that's where my, my head's at today, fellas.
Go ahead, Nico, go ahead.
I'm right there. And then like, and then a little one step further, I think, I think we're about to see the new era of, excuse me, the new era of like OPERATING system. But what's odd is that it's not going to come from the like base softwares itself. Right. So right now what I'm seeing is this shift from what a power user used. Like so it's, it's kind of funny because it's like power user was just the paid user before then it was like a max user. Then it's like, you know, max pro user or something like that. Right. And now what I'm finding is people who are just straight up like ap, like they just have regular monthly sub, but they're using the API. So like they're still burning like 150 to $200 a month, but it's just straight up API usage through like their own custom use of the model.
Right.
So I'm seeing, I'm starting to see this like next layer to your point where the context is, it's getting to such a point where like I don't think this is for everybody at all. I think this is, I'm just talking about like where I'm seeing things going at this point is that yes, it is getting to that like verbal orchestration layer where you're able to just now verbally orchestrate what it is that you need to have happen and once. And you're putting like this context layer underneath and that context layer where it used to be this, you know, data lake that you'd have to have, now it is just this underlying AI model that is connected to all of that stuff. And what I find super interesting and I sent you guys a link to it this morning. I'm definitely going to be checking it out today. Google just launched their cli.
Yeah, my ears and eyes perked up when I saw that come through. I want to watch that later as well.
Yeah, because that's because Claude Has. What is it, like, cowork? The Gemini cli is natively that way. So, like, how Open Claw works. That's Gemini cli.
Well, careful, because there's a lot of baggage that comes with openclaw.
Well, that's the thing. It's like if you took Gemini CLI and connect that to any vision model, or you just connect it to computer use.
Now you have openclaw without all the security risks.
Yes.
I'm just saying, watch the bag.
Yeah, no, I mean, yeah, definitely. I mean, there's a newer version of openclaw, by the way, that. That negates those risks.
The.
The biggest issue is that the initial version of it was trying to be like Moltbot as. As well and just like, really open source, which then created this, like, skill network. And the Skill network was flooded with skills that had, like, horrendous malware in them. So, yeah, be very careful with this kind of stuff. But, yeah, being the case where it's like, instead of having to have the. Where we were building the connectors, I was wondering where we're going to be in six months. And now where I see us in six months is you just have your own personal agent. Your personal agent goes around and collects the context for you. You don't have to. You don't have to do any of the connectors. The personal agent is your digital replica who just goes and says, like, oh, okay, I'll do this. I just need a key for this. It's just like when we're building out apps. When we're building out apps, it's the same process. We're telling it, hey, here's the prd. Have at it. Like, do what you need to. It goes at it, and then eventually it'll come back and be like, hey, okay, so do you have an API key for this or do you need. Do we have to set that up or. You know, as well. Right. And I see that being the case where, at least I hope that's the case, when I start messing around with Gemini, I don't expect to fire it up and, like, have full context on my PC right away, but the fact that it can, on its own, go look through the files and, like, I'm not going, Like, I'm expecting that I won't have to direct it to go look for stuff that it will just go look for it by reference, which is. That's what Aries does now. So if I tell Aries to go do something, go grab something from a folder, I don't even have to say what the folder name is. It has now built up the context that like, oh, he's talking about programming that's in his programming folder. I'm going to go there and go open it. Yeah, I mean it's, it's kind of like prompt engineering. That's why I've been like ignoring contextual engineering. I don't talk about context engineering because I think it's a problem that's about to fade into the background. And it's not like it's not important and it's not like it's not important to give your model context on what you're doing. I just think that very, very soon having to do that is not going to be a thing anymore.
Well, I don't know if I, I don't know if I agree with either of you, by the way, this morning. One context is definitely getting a lot easier. But just because it's getting a lot easier doesn't mean that the mindset of it gets less important. Right? So like is it going to be easier to put context in or is context going to be just automagical? Yes, but the way that you command, question, mend, bold, mo, whatever word you want to use there as a user, as a human, to the thing that you're doing to grab the right context from now, all the context is still going to be an important thing. Chris, I want to get back to you for a second because do I Want to code HubSpot CRM? No, nobody got time for that. There's no way. It's too massive. Do I want to be able to create things that once we're not able to be created and do I want to create things on top of other things? Let me explain. You mentioned CRMs. Let's mention CMSs. There's an entire project I've been working on that is on a completely different CMS Sanity IO because of the amount that I can layer on top of it easily with my AI assistant that I can't do currently in HubSpot CMS. Impossible. Okay. There are things that I can create and do that. For instance, over the weekend I wanted to put some courses together based on all the context I have of four cornerstones and 10 pillars for the Superhuman Framework for Faith Driven Leaders. There's a lot of context. There's literally six chapters of a book that are sitting there as well. Context. And to create the courses historically for those things would take masses of amounts of time and immediately we would go to a conversation of like thinkific or some course Platform. Now, will I recreate a course platform with AI? Absolutely. My AI assistant put together 21 courses with lessons and modules. And then what's crazy, it was like, hey, until I get the chance to record these videos, can we just give them very conversational audio using my 11 Labs integration for these so that we can launch these to the world? And the world don't have to wait on me, but they can have the text and the audio and I can come later and put the video in place. And my AI system was like, yes. So did I create a brand new platform to add value to the world? Yes. Would I do that with a CRM if I knew that I only needed 20% of what HubSpot provides?
Maybe.
Is that me? No. Is that somebody? Abso freaking lutely. So like, we gotta be careful because there's like the world that we live in and the positions that we have around these softwares and how much they're tied to our life. I've been in the HubSpot ecosystem system for 14 years. I'm not going nowhere. I'm gonna continue to teach the last human that decides that they still want to use HubSpot. HubSpot. But that's not everybody. And that's not every CRM or SAS or platform or thing. Like, it's getting wicked out here, brothers. Like, I'm just, I'm just saying, like, there's a lot to think about and there's a lot we can be doing. My biggest fear out of all of this is like, nobody cares and many people aren't doing much. That's my biggest fear.
Right. And that, like, that closed stance about don't do this thing. Right. Like that narrative does not support like opening people up. And there's a, It's. I've seen it in various consultants. Like, it's growing, obviously, like their livelihood is tied to helping people.
Yeah.
You know, with these tools. But that's why we say, like, usually HubSpot is not working. Has nothing to do with HubSpot. Yeah. And if we can enable people. Because guess what? HubSpot's never had a more solid place in my stack now that I can make it the universal context layer very easily because Claude code is there to manhandle. Or we need a different word for man.
Yeah, we need a different word to.
Dominate any API that is available.
Not a good word.
And so what if I told you guys that because I'm hearing this conversation, it's, it's interesting. What if I told you that there are currently agents that hire people.
To.
Do things for them.
Yeah, I think we should wait.
Let's get specifics. And this is like tomorrow's episode. Did I just hear you, Nico, say that there are AI agents hiring humans to do things?
To touch grass?
Yes, that's tomorrow's episode. Yeah, we have no time for that today.
Right. And I do want to like something that, that came up over the weekend, like as I was moving, right. And I keep coming back, George, you've heard me say this, like this idea of planning ahead, like for everything and trying to be prepared and managing risk and like all this possibility, like all the energy that we put into that. I was moving and I was like, man, I need to throw something away right now. I don't have the trash can yet. Okay, let's get the trash can on the next trip. Right. And all I put into my global Claude was just in time productivity, right. From my manufacturing background. Just in time manufacturing used to be a thing so that you didn't have lots of stock and it managed your resource risk in a way that companies did very well. His response. And by the way, this is to Nico's point where this is in global cloud without a project, all it has is the memory of us doing chats. And all I give it is in quotes. Just in time productivity. He comes back. That's a sharp frame, Chris. It flips the industrial age just in time manufacturing concept, which was about efficiency in production, into something about efficiency in attention and action. The implication being in an AI native world, you don't need to pre produce productivity systems, batcher planning, or stockpile workflows. You produce the work at the moment of relevance, which ties directly to your thinking that relevance requires timeliness. And it ended up being a new framework and I ended up moving from productivity because that's not really like, that's still very industrial thought. And what it can be easy to tie to is when in the Matrix she needs to fly a helicopter and she's like, let's figure out how to do that. So it's just in time learning for just in time execution. And this is like my favorite part. As I was going through this with Claude, he's like, oh, do you have a third thing for me? And I was like, yeah, nothing's coming to mind. But if you want to take a look at this book that I just started drafting and you know the website, I'm open to your thoughts. And this is like the. The brand being built right now is value first. And I did not come to this idea but he's like just in time value. And I was like, obviously like, wow, that's so perfect. And just this concept of the capability is right there. But all the energy we put in to building up for just in case situations of, you know, all of the planning, all of the risk management, I think as we're delivering, you know, production ready AI systems, I think that changes because we can now it's much easier for the humans to gain the capability to troubleshoot and to think and act in the moment, in moments that we cannot possibly plan for that actually create risk, like in process. When we can enable the internal teams to do that, we, we get to deliver in a different way. And instead of like AI just being brought in to handle everything, the delivery actually becomes like capability across the entire system which includes AI and humans. And I'm going to be like continuing to go down this path. But it is that, that, that's capability, democratization and access that we've never seen before. That is the change. And if you are trying to protect yourself by, you know, maybe suggesting that it's not a good idea to add that capability or to go down that path or just give it a try and see what happens, you'll find very quickly how hard it actually is to vibe code a CRM if you try to, if you try to do it yourself. And you will learn a ton along the way.
Well, and the amount that you have to know just to even start that project.
Right.
Like, do I feel like after 14 years of being in HubSpot, I could get close to not HubSpot, but close to a viable solution? Maybe like human is not going to be like, hi clock. Create a CRM. Beep bop, boop. No, no. Right, that's not happening.
Yeah, like let's, let's build a customer relationship management, anything with management like in it. That's not the ask folks. It's like how can we do this stuff? How can we keep track of this data that I need the next time I need it? But yeah, we'll keep going tomorrow. I am very interested to hear about both the Google CLI and whatever that other thing was we were going to start talking about and didn't. So until tomorrow.
Agents hiring humans. Oh yeah, bookmark that in your brain. All right, let's close this bad boy out.
That's worth the headline tomorrow. Have a great Tuesday everybody. We'll see you tomorrow.
