Latest / The Chief Customer Officer Podcast / EP033: Building Your Own UI, Named Agents, and a Bank's AI Playbook

Transcript

EP033: Building Your Own UI, Named Agents, and a Bank's AI Playbook · The Chief Customer Officer Podcast

• Jeff Breunsbach: All right, welcome back to another episode of the Chief Customer Officer Podcast. This will be coming out to you on Thursday, September seventeenth. Nailed the intro. Jay, how's it Jay Nathan: You got it right. Jeff Breunsbach: Jay, how's it going? Jay Nathan: It's going great, man. I I'll I'll talk about this podcast episode that I listened to over the weekend, I think it was. but the the one thing I wanted to tell you is that that podcast had a six and a half minute intro between like all the intros and the ads and everything, six and a half minutes. So if you don't know and you're listening to this, th first of all, thank you. But we pride ourselves on jumping right into the content so you don't have to waste your time hearing us banter too much. Except for now we're doing it a little bit. And you don't have to hear ads, right? So we do this for the love of the game. not for six and a half minutes of intro. So anyway, like Jeff Breunsbach: That's I mean that's that's kind of absurd. I was actually just thinking about Jay Nathan: Isn't that ridiculous? Jeff Breunsbach: like how long six and a half minutes is. Especially like I I don't even listen. I I'm pretty sure most everything, my YouTube, my Spotify is either 1.25 or 1.5 speed. so like six and a half minutes to me is now sounds very long. Like Yeah. Well, let's jump. Jay Nathan: Eternity And the whole thing was only thirty minutes. Jeff Breunsbach: Yeah. my gosh. Talk about getting your money's worth. Jay Nathan: Yeah, right. Jeff Breunsbach: okay, so let I want to show you I wanted to show you something. that I just whipped up last night. And I guess a couple of things that I'm like learning and getting into right now that I have found like super cool as I've been doing this. So you know, I'm currently building a platform for my team to use here at Junction, customer success platform, tying in all our tools. I've just decided to go kind of do that. And I think it's the right choice. I think there's a lot of like already like wins that I'm seeing, especially I've shown it to a few people. I I just showed it to you this morning. It's really not anything special yet, but I've already shown that to like our solutions engineer. I've showed it to our our solutions support engineer and a couple of other people. And and their ideas are already like you can start to see their wheels turning, like, okay, like centralizing this stuff. Like they can start to see, okay, like how useful this would be. And they're starting to even think, like, This could be a tool that actually like we actually we actually use outside of customer success. Like this could be a tool that like we actually like have multiple people like across the business. Our ops team, support team maybe logs into it. Like there's like true, you know, like value in this. so a couple of things that I'm a couple things I'm learning, and then I want to show you something that I did last night. And I I I literally like in my in my shock and awe, literally sat there for like three minutes being like I can't believe it just did this. So a couple of things I'm learning and playing with is I've I've been on Claude, I guess, like probably for the entire year, nine months. Like I've just been using Cloud Code. and we have codecs at work as well. And so now I've started to interplay them against each other, like tet like checking each other's work, Jay Nathan: Yeah. Jeff Breunsbach: testing out like ideas and theories and kind of like going back and forth. and so that's actually, I think, like been super helpful because I I think I almost like started to get into a mode of like, I just kinda trust Claude. Like, yeah, it's it's done great work for me. Like I'm kind of trusting it. And now that I'm doing it back and forth, I'm like, okay. Like this is actually very useful. We need we need checks and balances. and having and actually even something I've learned is you can have it, you can have different models check your work. And it actually will do very similar things too, right? Like it actually like has the same effect. But like I like doing codecs versus Claude. So that's like one thing that I've been doing more recently that I thought was super cool. the second thing Jay Nathan: How do you do that? How do you how are you physically doing it? Are you literally like taking codecs and pointing it at the same code base and saying, Hey, check check this and give it like a set of criteria? Jeff Breunsbach: Yeah. So yep. and I and what I'm doing now too is also opening up PRs and and then I'm at and then I'm going and having codecs like pull up the the last you know, the the PR number seventy or whatever. you know, read through it, go check it, tell me what you see. Is there anything that you should you would do differently? So then I'm kind of like asking it questions and probing it back and forth. and so that's just the way I've Jay Nathan: Okay. And Jeff Breunsbach: been doing it right now. Jay Nathan: and pull PR is pull requ pull request, right? Jeff Breunsbach: Or quest. Yep. Jay Nathan: Which is like a it's like a unit of it's like a a release, right? Jeff Breunsbach: Yes. Jay Nathan: Or like a like a unit of work that you've done on your on your platform. Cool. Okay. I love it. Jeff Breunsbach: Yep. so and then the other thing, I just lost my train of thought. Good for a podcast. No no, it's okay. Jay Nathan: Sorry, I I I did that to you. You're turning into an engineer, by the way, while you're thinking, like you're becoming an engineer. Jeff Breunsbach: It's yeah, it's kind of fun. I mean it's super it's cool that I mean like being a builder, like being able to build something and then use it and see it, like there's I don't know, it's like there's like a reward mechanism. Yeah. Jay Nathan: You've always been a builder. You've always been a builder too. Yeah, you're getting dopamine hits all day now. Jeff Breunsbach: Yeah. unless my tokens run out and then I'm you know, what give me more give Jay Nathan: Yeah. Jeff Breunsbach: me more tokens. That's what everyone's saying. Okay. so checking Claude back and forth and then what was the other? Okay. I'll I'll remember it at some point. Okay. So I am so I'm I'm building a section of the platform that I feel very passionate about and I think like needs I think a level of like I guess precision or like to the highest degree. So I went into Fable last night and I said go pull the code base and I want you to build me a mock-up. And I gave it like other systems that I like. So I'm I'm building kind of a project management system, initiatives, projects, tasks, and think about things that you might want to do with that, right? Like, hey, I want to be able to I want to be able to roll out an initiative to multiple customers, but I want those projects to show up, you know, on those customer records. I want to be able to like ladder those up and ladder those down. Like there's some complexity there. And so, like, you know, there's all this kind of functionality that I want to make sure is like built in that we can do. And again, this is going through HubSpot's projects and tasks data model is like the under undercarriage. And so, like, you know, still using a sort of a system of record, a source, you know, source of truth as like the the base that we're working from. but I told it, I said, Hey, I really enjoy linear, like I like linear's like of their modal pop-ups and some of this other stuff. I gave it some other systems Jay Nathan: Yeah. Jeff Breunsbach: and said, Hey, I like some of the look and feel of this. And I said, here's what I'm trying to accomplish. And then I asked it to like read it back to me. And then I said, okay, go build the mock-up. And first shot, the mock-up that it built. man, now I'm sharing a screen that I don't even know. So the what it came back with is a Claude artifact that is seven different screens for a project management tool. Jay Nathan: my gosh. Jeff Breunsbach: And it gave me like all the model, like all the data fields that it was using, what what models those roll into in HubSpot, like you know, how this would work. But this is essentially like what it came back with. So screen number one that you're looking at here is like here's what initiatives will look like. So now I can ladder up and ladder down initiatives. again, if you kind of look at this, like some of the things I talked about are built in. Like, okay, how many projects do I have under those initiatives? Which accounts are those tied to? Who's the owner? What's the priority? Tagging systems, right? So like you kind of see this first one. Like to me, this is like. Exactly what I need, right? I the I can ladder stuff up and down. I can like dig into stuff. This is like a very high level view. Okay, cool. Well now let's dig into like what does an initiative record look like? Great. I want to be able to see what those projects are if I click into an initiative. I want to see which ones need attention. I want to like look at activity. you know, see some details about this. So again, this is kind of great. Initiative, you know, record is done. Okay, now what if I dig into a project? What does a project look like? Okay. Here's the project you know, here's the task list that's going on. And again, this is still at like a you know high level. but like here's the tasks that are going on. Here's the again, all the details about this project, open tasks, anything that's like related that we can tie it to. Then it goes to your accounts. Okay, so now I can see these projects tied on an account level. So this is my account now that I can see for Northwind Health, which is this made up company. again, like how much ARR and everything, but this is the projects tab. So now it again kind of shows you in context. now I also have a board. I can look at these projects across the board and like what's happening with them. Now I have a timeline that I can look at, you know, and see how these things like start fitting together. And then here's Jay Nathan: Yeah. Jeff Breunsbach: what the modal looks like in terms of like a task dialogue, if I'm gonna add a task or whatnot. And so again, I just I I guess like continue to be like shocked in some of these cases where like this this was literally I could like show you the prompt. It was literally like a throwaway prompt where I was like, hey, I want to build this project management thing. in a part of the tool, go grab my code base. Here's some examples. And like this Jay Nathan: Yeah. Jeff Breunsbach: is what it went and built. And I just kind of like walked away with like, wow, like that, that's a pretty wow experience to me of like, okay, it built seven screens. Largely all the functionality that I want is there because it, you know, was looking at systems or or tools that I I told it that I liked. I also went and found like a GitHub repo for like an open source project management tool and said, like, hey, You know, let's look at this to see if there's things that are are useful. and so like, you know, just I don't know. I I just wanted to show you because I thought it was so cool that like literally in one shot it did this. And then I said, okay, now I want you to like I want all of these things verbatim. How build me the plan to go do it. And it built me literally an entire page. I had that, checked it against codex and basically like ironed out this plan. And so then it came out with seven different. There's seven screens. So it calls, you know, it's S1 to S seven. And then it built a multi-agent approach. And I just it just finished right now. And it like, so it just went and ran multiple agents. And again, kind of creating PRs, getting codecs to review these things. Like, and now, you know, hopefully after I get off this podcast, I can go look at something that's like actually working. Jay Nathan: You can't wait. You can't wait, can you? Jeff Breunsbach: Yeah. Jay Nathan: all right. So let's take it up a level because I mean, first of all, like this is amazing. Second of all, shouldn't be surprising, right? You know, you already know that Claude can do this, right? It just did a whole bunch of stuff at once, built built an entire application off of one prompt for you, which is cool. But it it ties to what so one of the things I wanted to just touch on was the the Dreamforce keynote yesterday. And they are like, they have made a very big bet at Salesforce that they don't the their user interface is now AI, an AI agent, right? So it's Claude Jeff Breunsbach: Mm-hmm. Jay Nathan: or in fact they had Dario at the he's so awkward by the way. He's such a weirdo. But they had him at the keynote. I'm probably going to get murdered now that I said that on the podcast. but He was there talking in in the whole idea is you either use Salesforce through Cloud or you build your own interface. And I've actually seen this with sales directors, regional VPs, A VPs, they are building their own views of the CRM data. So what I would contend is that you're not building a platform. You are building a UI. And I think the the in it's exactly what you need. Right, which is which is fantastic. But the platform consists of the data layer, which I think still needs to be codified for you. It's in HubSpot. You know, for the people on AI force, it's gonna be Salesforce. And the so I I actually think the next frontier, the lovables, the base forty-fours, the replets that build these UI layers on top of data platforms are going to get even more valuable because they really do need to be integrated into these into these platforms, right? The like you could imagine having Claude Claude or or Chat GPT or whatever the platform is having a, you know, an interface's area. Right, where you actually told it what you wanted to connect to and what you wanted to build, and it's gonna build that, and then it's durable. And then so I think the the the DevOps, the RevOps, not the DevOps, sorry, the RevOps of the future, you know, instead of they're of course going to be designing agents and workflows for the more deterministic things, but they're also going to be generating user interfaces because not everybody's gonna go do it. And not everybody should have to do it, right? You should be able to share interfaces across these applications. so I think it's really cool, Jeff. And you know, I think it y one of the other pieces of it that's really I think key is like all the permissions and everything. St you still need a platform layer for that. That's Jeff Breunsbach: Yeah. Jay Nathan: not gonna ever be managed by the model, I don't think. you need sort of a a very crisp, clear definition of who's who's able to access what components and elements and update this and view that like that's all still very granular. That's where that's where the CRM comes in. Now the question is how much will that be worth in the future? Jeff Breunsbach: Well and that's like Jay Nathan: I don't know. Jeff Breunsbach: one one of the things that I feel like I'm moving a bit slower on this. because I'm being so I'm trying to think of the right word. I'm I I wanna try to build this in in a way that like you said, like has has the thought of permissions and controls like baked in from the beginning, right? Because What I can foresee is I build something that's so useful right now, but then like two weeks from now, I'm like, well, I did not think about you know, controls at a field level. And now one of my, you know, somebody messed up a field and it went back into the CRM, and that's a field we should never touch, right? So cool, we should be doing field level permissions right now. and then we should be thinking about, like you said, we should be thinking about views, okay, or like kind of seats, right? Like who's who's got like certain access to like open up. you know, and be able to do certain things on certain records. And like I guess that level or attention to detail is probably something I don't I I could probably move faster and not think about it, but I just think if I do it now, like I'm not gonna have like I I'm just going to have to maintain it as I go forward, not have to like rebuild something or like totally think about the architecture differently. so that's like that's one thing. and I think, you know, to your point, like What I I've shown I just showed this to like our sales leader and he was like, dude, this looks so nice. And I'm like, Yeah, like it all it is is just like HubSpot data. He's Yeah, but it looks so like I like I want to use it versus like what like HubSpot Jay Nathan: Yeah. Jeff Breunsbach: right now, like I don't that that interface like by the way, talk about a place that you go to like click a ton of buttons and like nothing happens and like you just like keep clicking. It's like Why do have to click into a deal record that opens up a whole new page that now I'm on like a different deal record? I can't tell if these activities are to the deal or to the main page. Like anyways, just confusing. But yeah. Jay Nathan: matter though, luckily. I mean you can you can do it, you don't have to anymore, which is great. I agree. Jeff Breunsbach: Yeah. and then I guess like the other thing that this like you said, the other thing that this continues to call out for me is that is like your your point, which is like we've got to start, I guess like getting our teams and coaching our teams to like think of okay, can I go get like some can I go get an agent to do this first? Or can I go get not even an agent, like cowork or code or claude work or what's it called? Chat GPT work. but like, can I go get it to do it, do this thing for me, versus having to go do it. And I still think that like everyone is is in the the mode of like, okay, I need to go find this project and this task and click this button and change this and whatever, versus like, hey, you Jay Nathan: Yep. Yep. Jeff Breunsbach: should just like prompt it to go do it. Like, yes, I like I still think there's value like in having a UI. Okay, I can I need to be able to see things, understand maybe dependencies and how things trickle down and like what's happening, but like It still doesn't mean you need to go to the UI for everything that you do. and so, you know, I think that's like something. The second thing that I'm like coming out of this with as well is just the the idea of of working where where your team is. I I don't this is clearly not a new new concept, but like we are so slack driven that like right now, as I'm developing this, I'm sitting here thinking, like, doesn't matter how cool this tool is, doesn't matter anything if my team is still just gonna be in Slack like eighty percent of the time and not be able to use this. Like we need to have a Slack bot that can be called upon or like clawed that's connected in. Like there's gotta be some way that they can essentially like call customer OS and like do any of the things that they want to do that they normally would. Jay Nathan: Okay, so let me let me tell you the way that we're sort of thinking about agents right now and and building agents. So if you have like you had agents in your in your application, right? I would argue that those are probably more like automations. They're things that happen automatically when a call gets recorded or when a deal record changes some of the stuff that you mentioned to me before the call. Agents are like employees. So what we started doing. is instead of a Slack bot, we're just gonna we've we've provisioned agent our our our agents an email address and a Slack account. So they're just like another employee. And they're gonna have a name, right? And there are certain tasks that we say, hey, you know, John, our our agent for like administrative stuff is going to handle this particular task and we scope that task to John. Right. So we just treat him exactly like that. But there necessarily, I think, to your point, you want him to have an email address. You want him to have a Slack account because you want to interact with him in those environments. You want to bring the AI to where people already live. And I think, I think you're spot on there. Like that's where people live. Jeff Breunsbach: Yeah. Jay Nathan: I talked last week about T Town. such a great Such a great agent. You know, we'll we'll see. I mean, Instinct is another one, by the way. They're, I guess, trying to raise it ten billion dollars now. We'll see. We'll see. Everybody's building personal agents now. So it's gonna be interesting to see who actually wins that fight. But it's the same concept, right? Just just bring intelligence to where I am and you Jeff Breunsbach: Yeah. Jay Nathan: know, we'll see if it we'll see if it works. So Jeff Breunsbach: Muse Muse just came out too. I feel like we have people a couple of people at work who've like already played with Muse and like Jay Nathan: Is that meta? Jeff Breunsbach: Yeah. And people are already saying how like it's like one of those things where people are kind of counting meta out and we're just like, you know, you should never count out Zuck because like I think they just ran like testing on it and it's like all the, you know, across all the I don't know, variation of of tests and the and it's come back and passed h higher percentages than like any of the other models that are out there right now. and like I had somebody Jay Nathan: Wait, is Muse the agent or is Muse the model? I don't I don't haven't kept up with with Meta's Jeff Breunsbach: I'm confusing myself. it's Meta's personal AI agent. Jay Nathan: Okay. All right. Jeff Breunsbach: so like one guy had an example where he like scanned a citation and said, Hey, I need you to go pay this. And it like found the website, you know, found like put in the ticket number, literally had all his login stuff, like and just like went and did all this stuff for him and said, Yep, you know, went and paid the ticket. And all it was was just he just scanned the ticket. That was it. Like he didn't give it like anything else. and so there's some examples like that. this is also what I think is what I've like found interesting too is have you heard of Lightfield? Jay Nathan: Mm, no, I don't think so. Jeff Breunsbach: so Lightfield just came out and it's kind of like it's gotten a a relatively amount of hype on like Twitter or LinkedIn or somewhere. I read about it. And it's like this agent or it's this agentic CRM is like the way they're positioning it. But what I think is super interesting about what they did is one of the things that they're doing is like call recording, for instance, is actually natively built into the platform. So you use their call recorder and then it goes and does actions in the CR like. then it's basically built off the back of that that it goes and does actions for you. Which like again, you think about these call recorders that are out there now, Gong, Fathom, like they have not taken Jay Nathan: Yeah, yeah. Jeff Breunsbach: this approach, right? Like they've just been, yeah, we're a call recorder. Like we're gonna get really good at call recording. And like we're gonna give you the task list. We're gonna give you all the stuff, right? Like, and you know, at least from my, you know, little view of their website, I'm and I I'm like, like they they're like, no, let us not just record this. Let us like actually then just take that call you just had and we will go do everything in the CRM for you. Jay Nathan: Yeah. Jeff Breunsbach: and like it's natively built in. And I just think that's a like to me, do the like do the I think we've like probably assumed this, but like do these little features, these little WYSIWIGs that have not been platforms, do they just start getting eaten up by platforms who like just build the feature natively into it. Right. So like Why would you ever use gong if Salesforce had a great call recorder that automatically worked with Slack Jay Nathan: Yeah. Jeff Breunsbach: and then went and did all the stuff for you? Right. Like Jay Nathan: Well, I ask myself that all the time. Like we we have Zoom, we have Fathom, we have I think we have another tool that does call reporting. I'm like, should we actually be paying six thousand dollars a year for call recording through Fathom? Right? Because we get it for free. Jeff Breunsbach: Probably not. Yeah. Jay Nathan: Transcripts and everything. We just have too so much hooked up to it now. It's like, okay, well, and people know how Jeff Breunsbach: Yeah. Jay Nathan: to use it. Zoom's usability is pretty rough, but I'm guessing you could probably act I mean the reality is we log into are fathomed less and less because it's all accessible via Claude now. Jeff Breunsbach: Yeah. Jay Nathan: So I think you're right. I did see the light field thing. I I mean now now that you you jog my memory on it and I I remember sort of rolling my eyes at like we're you're re rebuilding the CRM from the ground up with agents. Like yes, I mean that's fair, right? But what can they do? What can they do that that HubSpot plus Claude can't do. I would argue you still need a tr a structured data set behind the scenes, right? A cost aside, right? It may Jeff Breunsbach: Yeah, yeah. Jay Nathan: be the fact that they can combine the the token usage with the data model and get a much more efficient, it's like a vertically integrated solution, right? Where it's much more efficient because they know exactly how to work with it. It's not like it's not like a PC, it's more like a Mac, right? Everything's integrated from the data model all the way up through the the model, like the intelligence model layer. But like do I actually need does is that necessary? I think they're it's so easy. Like look look look what you built in a week. Like it's it's actually not that hard to do what they're doing from my perspective. Jeff Breunsbach: Yeah. But I I guess like Yeah. I think like to me though, like the like the more interesting question is like, okay, the companies that get started today, right? Like if I was to go start a company and we're growing and we're ten, twenty million dollars, I just raised fifty million dollars, right? Is that person choosing Salesforce and HubSpot today? Or are they looking at them and saying, you know what? I could actually am Jay Nathan: Great question. Jeff Breunsbach: I am I picking, you know, Salesforce and HubSpot and then Fathom or gong and then like stitching all these tools together or are they going and looking at a light field saying, this is kind of interesting. Like I can just give my sales team a call recorder that just does all the stuff for them. And then it's in a system of record that now I can build around that has probably like like more up to date API k APIs and protocols and and you know, whatever else. I guess like that's the more interesting part to me. Like you said, like I don't think there's, you know, is Lightfield gonna get a Fortune five hundred company to switch from Salesforce or HubSpot? Like, no. But like Jay Nathan: No. Jeff Breunsbach: Does the new like do these new AI forward companies start looking at the clock start looking at the Salesforce and HubSpots of the world as like, okay, that's like old, like that was built, right? That was built without AI and agents ever in mind. This new light field thing is built with AI and agents in mind. and so I don't know. That I guess that's the part that I think is interesting. Is like that to me is like, I don't know, are they gonna win? You know, HubSpot's and and Salesforce have won the last 20 years of. companies just because of their name recognition and okay, they've been in the space, you know, it's like it's kind of like buying IBM, right? Like I'm not gonna get fired for getting Salesforce or HubSpot in here. And there's plenty of partner ecosystem, there's plenty of all this stuff. But now in this new frontier, do they win the next 20 years of of new companies starting, which are going to then be multi-billion dollars, Jay Nathan: No. Yeah. Jeff Breunsbach: multi, you know, whatever. Like that's the interesting thing. Jay Nathan: Yeah, yeah. I think you're I think you're right. I think you're right. Like you've got to win the startup ecosystem ten years ahead of becoming HubSpot or Salesforce, right? But yeah, no, I mean no no unless you're like very well funded and you know you're going straight to a hundred million in year one, you're probably not taking Salesforce off the shelf. You might be using HubSpot. So I I buy it. I I think the I think the use case on that though is a little bit more practical. Though it's like, hey, call recording, you pay X amount a month for that. And CRM, you pay X amount of month that. Like I look at both of our our CRM and call recording spend, and then not even to mention Zoom. And I'm like, well, okay, like it may be maybe you know, no different if if I just move over there and have everything in one platform. So I think that's the other thing is these quote unquote compound platforms. Or compound startups where they have multiple products being launched. I used to think that was sort of a I was like, okay, like that's a that's a little bit of a contrived concept before AI. So I looked at what Parker Conrad was doing with Rippling, and he's been talking about compound startup for a long time, right? Like we're building a very wide product. Well, Rippling raised 1.2 or 1.3 billion dollars. Of course, you better build a a platform for that kind of money, right? But in the age of AI, you don't look what you were able to do in a week, right? You don't have to build or you don't have to raise that kind of money to go build a compound startup. Like every startup should be doing something. The other thing I think would be an interesting way to look at this is okay, what is the light field for consulting firms like mine? What is the light field for software firms like Junction? What is the light field for industrial services and manufacturing firms that have heavy CPQ requirements, like that you don't want to go bolt on a CPQ. The CPQ is the product, right? So I I think that becomes way more interesting as well. So I don't know if there is another Salesforce ever in the future or another HubSpot ever in the future. Could one even ever get that big because they're going to be fragmented solutions that can serve the needs specifically. of an industry or a particular domain. Jeff Breunsbach: Yeah. I think it that is like the I think the talk, right? Of like, okay, now if I can configure, customize the solution to fit very niche industries, but those industries I can go deep and have like this depth of knowledge that like is super hard. Like you said, like CPQ is an example of like, okay, if I just became the best at CPQ, like, you know, would this then, you know, be the like the right thing to do? And you could probably go build like you said, I wonder if it if if it starts to spark like a I don't know, more modest size software companies, right? That end up being like they Jay Nathan: that's what I was thinking. Yeah. Jeff Breunsbach: end up being 20 to 80 million dollars, and you know, they can cash flow and be positive and not have to, you know, it's it is kind of interesting. Like the the other thing I was watching invest like the or not invest like the best, I was watching founders with David Center. He just had the founder of Betting Spoons on there. And Jay Nathan: Bending spoons. I have a rant on that one, by the way. Jeff Breunsbach: What's super interesting though is that this is like you do think about how software companies have this is a large generalization, but a large portion of software companies have like looked at this model and said, Okay, I'm either zero or I'm a billion dollars, right? Like it's I've got to I've gotta pick a lane and go. And then you look at like the thirty seven signals of the world who are like, We're bootstrappers, we make a twenty million dollar business and cash flows and we just live our life and this is great. And like, like, you know. That has been almost like apropos or like looked down upon almost and like it feels like for me, people have kind of said, like, like that's a whatever, that's a small business, you know, that's you know, that's not my billion dollars over here. But I wonder now, like you like you're alluding to, like, okay, is it like way more accessible for me to build a 10 to 20 million dollar software business for even Charleston, South Carolina? Could I go build a you know, a 10 million dollar business for South Carolina? something in South Carolina, you know, that like and like could that be something? Like, I don't know, 10 to 20 million dollars might be a stretch. But like it just seems far more likely that like the scenarios of zero or a billion dollars actually comes down to like, you could build 10 to 10 to 30 million dollar business pretty quickly and like very cost effective, cost efficient, and probably cash flow positive because I don't need as many software engineers to actually get an MVP like I could go scope an MVP. I could go scope a a product pretty quickly now with FOD. And like that to me is just seems like way more accessible than it ever has. Jay Nathan: Hundred percent. That bending spoons episode, I haven't watched the whole thing yet, but I saw a highlight where he's talking about not having titles. Did you hear that part? Jeff Breunsbach: I d I haven't gotten to that part yet. I haven't it's kind of a tough watch. I I find his his style or like his communication, I don't know. It's it's like Yeah. Yeah, I think something that. Jay Nathan: The the founder or the the Ferrari guy? Is that his name? Ferrari. Cool name. Jeff Breunsbach: but so I haven't got to that part yet. What is it though? Jay Nathan: Okay. Well, I just well, I mean, they're basically I have to admit I didn't even watch the the clip, so I probably need to go listen to the whole thing. But David Center's post on it said it, you know, they're doing something so innovative. They're basically they tell people you don't have a title. Like you can go whatever title you want, just don't embarrass us. Like put it on put it on your LinkedIn, like you can have whatever title you want here at Bending Spoons or at a Bending Spoons company. And I got to thinking about that. I was like, you know what, for millennia, humans have structured their tribes, their groups, with different titles, right? Like you you you have to know what the hierarchy is to some degree and and what the role people play is. I just think it's one of these like I think it's a concept that a company that's gonna own a business for three to five years can come in and just blow it up and be like, Yeah, whatever. F titles, it doesn't they don't matter. These people know nothing about management, right? About leading teams to an outcome. And so I just I don't know. I think it's I thought it was really silly, honestly. That that's the approach that they take. I was a little surprised by that because I thought a little bit more bending spoons, honestly, before that I Jeff Breunsbach: Yeah. Jay Nathan: heard that. I thought it was stupid. Jeff Breunsbach: one of the things that I I did think was interesting, that I think he talked about, the CEO of of Betting Spoons on there, is he kind of talked about how the concept of a a you know, I forget, Evernote's been around for what, twenty years now, maybe? Jay Nathan: Yeah, one of one of my investors in a former company on our board was in Evernote and they couldn't get it to grow back then. It was a real problem. Yeah. Jeff Breunsbach: That's funny. but he said, you know, like the concept of going to work for like an Evernote now in 2026 is like kind of a hard, it's a hard pitch to like, hey, you're a maybe you're an and coming software engineer. You want to come, you know, try for the 20th time to revive this user base and like what do, you know, whatever, like work at Evernote. And so he said, like, what's what he's what has been interesting about like almost like pitching bending spoons to somebody is hey. We're acquiring these companies. And it's not that you have to work on that company for five, 10 years anymore. You know, like you could go work on a project at that company for a year and then move to the next air table. Then you can move to the next, like you could write in Jay Nathan: Yeah. Jeff Breunsbach: like I think what his idea was though is just like, the concept of now being able to like I guess like not be stuck at a company like that that's already 20 years old, that doesn't feel like it's got growth, that doesn't feel like it's going anywhere. Like now is kind of flipped because it's like, hey, we have four to five of those. And like the mandate isn't, you know, like, let's go change this WYSIWYG. He's like the mandate now is like Evernote hasn't been growing for 15 years. That means we have to go re-architect like the entire thing. And like that means monetization and like these funnels and like all this stuff. So now the problem becomes super interesting. And hey, we can work on that for 12 months. And then if you, you know, like now we've got this other one that's super interesting over here, that's like a totally different problem. And so, like, I don't know, this the whole conglomerate. idea. I forget the company that's based out of Canada where the guy's been doing this, like buying software companies. Like it starts with an E. Honestly. Jay Nathan: Yeah, yeah. I can't rem I'll think of it in a minute. Just small, right? Small cash flowing software companies. Yeah, yeah. Hundreds. Hundreds. Yeah. Jeff Breunsbach: but anyways, I I just Yeah, but he but he's he's like stacked, you know, multiple of them on top of each other. And like it just reminds me that like I don't know, there is I I do think there is something like to your point, and I think like what you were alluding to, right? Like the width of like this now like grows. And I think there's just like an interesting aspect around okay, like I get to work on interesting problems at these companies that like again, like we've acquired them because of like. their poor growth and there's not like much there. And so like it isn't just come in and let me fix this one little thing. Right. It's literally like, okay, we get to look at this like with a fresh set of eyes. And I don't know. I think it's kind of an interesting approach that he's had. Jay Nathan: But I wonder I wonder how how that actually works practically. Do you actually think that people can s just switch from one company to the other that seamlessly? I mean, the reason these companies got bought by Bendings by the way, just s if people don't know, Bending Spoons is a private equity firm out of Italy, right, Jeff? And they have been buying up sort of these upside down distressed assets. So they just bought Airtable. Airtable was a seven billion dollar valuation. They bought it for one at their last race, they bought it for one point two, one point three billion. They just did the same thing with Miro, Evernote, which you alluded to. they will do it with many more. They have deep pockets behind them. So, but I mean, these companies are are struggling. Their their financials are not clean. They're gonna be laying off half of the employees at these businesses. And they're not shy about saying that. Clearly they're gonna let you use choose your own title if you choose to stay. Congrats. Like you can give yourself a promotion, whatever. I and so, but like how how realistic is it to think that, okay, I I'm on Evernote now. I just want to go work on I want to go work on air table. So I'm gonna just go do that. I don't think that's very realistic in practice. I'd like to I'd like to know how much of that's actually happening. Jeff Breunsbach: Yeah. I think it's I I think it probably I think it happens and I think it probably operates a little bit more like a almost like a consulting firm where it's like, okay, I'm on this project, it's a twelve month project and like you know, there's like checks and balances. Okay, like if you want to move over to this project, there's gotta be somebody to take your role or there's gotta be some right. But like but I do think like that hap I I would a hundred percent think that happens. Like just think about a consulting firm, right? Hey, this is a project that we're coming in to revamp this and like great, I'm working on this for now, I'm working on this next one for later. Like Jay Nathan: Yeah, Jeff Breunsbach: I actually think Jay Nathan: maybe. Jeff Breunsbach: I actually think that happens a lot. And I think that like you said, like a who knows, you know, what how many more companies or whatever they buy, who knows if they're able to turn these things around. But I think I think it's an interesting premise in scooping up these assets like at a time when Jay Nathan: Yeah. Jeff Breunsbach: they they feel like they have no like utter like they ha kind of have no future value right now, right? Or at least like little. I guess they're they're valuable to the cash that they have right now, cash and cash flow. and I I just think like there's an inter it'll be interesting to see what happens in the next like five years with these companies. Like do they actually Jay Nathan: Yeah. Jeff Breunsbach: turn around and can they make them growing again? just because you feel like the Jay Nathan: Hey Jeff Breunsbach: the money tree ran out, right? Jay Nathan: Yeah, like it you know, one man's trash is another man's treasure and what doesn't it the As long as the business is not burning a ton of cash, which they're gonna solve that, right? You know, these are actually decent businesses with five, seven, ten years of cash flow ahead of them. And that's how you measure the value of the business. So if you don't overpay for it, like happened in twenty twenty and twenty twenty one with a lot of these companies, then it's a great asset. So, you know, they've they've really honed in on a on an interesting model. So kudos to them. Still don't agree with the title thing. I obviously I can't let that go, but Jeff Breunsbach: all right, the last thing I wanted to ask you about is have you ever used superhuman? Like the Jay Nathan: The email client, Jeff Breunsbach: yeah. Jay Nathan: I tried. I tried twice and I failed. I couldn't it just wasn't for me. Jeff Breunsbach: I I can't I can't get my head around it. Like I I and I just I don't know, like it's isn't for me either. Yeah. So what I got interested I got interested in this Jay Nathan: I know people who love it. I know people who love it. There were a client of ours at one of my companies. Jeff Breunsbach: for a minute. Did you did you see who Hu superhuman just acquired? Jay Nathan: No, I don't think. What? Are you kidding me? Jeff Breunsbach: Fathom. So they just they just acquired Fathom. no, like a couple days ago. Just happened. Jay Nathan: great. So it's only gonna be keyboard shortcuts for Fathom now. Jeff Breunsbach: But they so I I just think this is Jay Nathan: I didn't know that. Wow. Jeff Breunsbach: ending up being the weirdest company that I'm thinking of. Because if you think about so superhuman started as like an email client. did you know that they also like Coda? Do you remember Coda? It was kind of like a Notion competitor, kind of like build your own Jay Nathan: Okay. Jeff Breunsbach: pages, had like integrations, kind of like a style. Okay. So Coda Jay Nathan: Vaguely, vaguely, yeah. Okay. Jeff Breunsbach: acquired Grammarly. And and superhuman, like in one go, or like somehow they all came together basically, and then they all rebranded this under superhuman. And so now I'm like, okay, first of all, I don't get this brand. Second of all, what is this what is this amalgamation of tools that you're just like collecting? Like, why does Grammarly like sure, does Grammarly matter to email, I guess, maybe, but like, but now I'm like, okay, what does Coda have to do in this whole suite of things? And like now they're acquiring Fathom. And I'm just like, what I I can't from the outside understand what this strategy is. And it's just I just I went Jay Nathan: Ha ha ha. Jeff Breunsbach: down this little wormhole the other day because like we have superhuman accounts and I just don't use it and I'm probably gonna cancel it. And and then I was like, wait a minute, they acquired these tools and like what is this product strategy? I went and looked at their website, I was like super confused. I thought I go to their website, their homepage. I'm like Jay Nathan: Super confused. Jeff Breunsbach: and I'm just like, okay, are they building a platform to try and like compete with like like Google and like like Microsoft like, you know, like okay, come here Jay Nathan: Yeah. Jeff Breunsbach: for like your whole work I guess platform. And I'm just like, I don't I don't know. I it just I just left feeling super confused and I guess better about myself. I'm like, I at least I don't work there right now because it seems confusing. Jay Nathan: I love that you keep saying super confused. I just think it's such a funny thing to say about superhuman. It's nope, just super confused. Jeff Breunsbach: Yeah. Jay Nathan: but people people who like superhuman love superhuman. Like the guys Jeff Breunsbach: Yeah. Jay Nathan: that I used to work with at at the last company I was ad advising, just I mean, they're startup guys through and through. Love, love, love superhuman. Swear by it. And they became a Jeff Breunsbach: Yeah. Jay Nathan: customer. They became a customer of that company too. So Yeah, I had not heard all that. So I I I have not kept up with it. But that's that's pretty funny. Pretty funny. Jeff Breunsbach: Yeah, that's my fun story for you. Jay Nathan: Super confused. All right. I got I got one more thing. I got one more thing for you. do you have time to keep going for a few more minutes? Jeff Breunsbach: Yeah. Yeah. Jay Nathan: So I I wanted to share a LinkedIn post that I that I made and and really it's that podcast that I mentioned at the top of the session that I listened to. This guy named John Cowall. who's the CTO of a bank called P Pac Private Bank. I think they're out of the PAC. Jeff Breunsbach: Pack? What is this bank? Jay Nathan: I don't know. I don't know. But I'm gonna meet this guy and I'm gonna talk to him because I thought he was really interesting. But he was he he had such a pragmatic approach. Like think about a bank adopting AI, right? That's just a it's tough. Now this is a small bank. I think there's 700 employees, you know, sort of specialty. But still a bank, right? Regulated industry. Like everybody that I've talked to in banking is very scared of AI. very scared. And we have clients in in banking. So, you know, we we have some of these conversations. So but I thought his approach was very pragmatic and I wanted to outline it. So I I wrote this in in the LinkedIn post. You can see it on my on my profile if you want to go look at it, look it up. But I'll walk through the the elements of this. So First of all, the prerequisites, they built a data warehouse five years ago. And he said without that, this thing would be really, really hard to deploy. Agents would be really, really hard to deploy within the bank. So, you know, we've talked a lot about, you know, bringing data bet data sets together. You've got all these disparate tools, but like what is your core company data? Having that master data, you know, in one place, you know, very, very important. So they did that. And of course, prior to that, they had moved to the cloud. Duh, right. But that that's great. So they've got a data warehouse. tools and training. Three years ago, they gave everybody access to it was it was co the early version of Copilot at the time, Microsoft Copilot. So they gave everybody access to it and they gave everybody training. and that was everybody across the company. Jeff, I swear to you, I still talk to people that are in SaaS companies that are like, yeah, we they haven't given us Claude yet or Coworker, you know, Chat GPT. They Jeff Breunsbach: That can't be true. Jay Nathan: haven't given that is true. I I'm I'm telling you, there's not many now. There's not many, but Jeff Breunsbach: Yeah. Jay Nathan: there are some. They're like, no, we just we don't have that. I have to use my phone at work if I want to, you know, chat with an LLM. So, so that's a thing. So tools and training. the third thing is the CEO said, actually, I'm gonna skip one and talk about this first. The CEO said to the CE, CTO, like this is so consequential. This technology is amazing, but we wanna we really need to drive adoption of it. And so I want you, CTO, John, you need to be spending 70% of your time on this. So John's John basically took on a 70% role of adoption of AI and proliferation of AI across the business. as part of that, he formed a champions team. Champions team. Basically had people from every line of business representatives who were sounded like they were hand selected, right? From people who raised their hand. I almost said a like interview for this. They are all dotted line to the CTO and they're responsible for surfacing project ideas from their line of business. They know the business the best, right? They get trained on AI, and it's like, okay, now how do we apply this to my part of the business, whether it's underwriting, whether it's you know, in High net worth, you know, relationship management, or maybe it's the contact center, but those people know their business the best and they know how how it needs to transform. So they created this champions program. Champions meet once a month. They have one-on-ones with the CTO to review all their projects. So he coaches them and but he really pushes the projects out to them, right? They're responsible Jeff Breunsbach: Yeah. Jay Nathan: for building them. And then the the IT organization is it sounded like is almost like a support organization for them, right? It's like, hey, like. Go until you get stuck, and then let's figure out how we can how we can help you get unstuck. I just love the the the culture that they built around Jeff Breunsbach: Yeah. Jay Nathan: this. and then, so then named agents are really how they've deployed agents. And that's back to what you and I talked about a little bit earlier. Every agent is like, hey, this is Sally. Here's what Sally's job is at the bank, right? That helps you scope the context for that agent, what it has access to. and here are the here are the tasks we're gonna assign Sally. So we got Sally, we got John, we got Jeff, we got Jay. These are our agents, right? They have personalities and here are their responsibilities and roles so that you can interact with them on Slack, email, and it goes back to the town thing. I keep talking about it, but it's like that's that was such a it was it was a good revelation for me to just interact with town and see like, okay, this is what it looks like when it meets me where I am. And it does a Jeff Breunsbach: Yeah. Jay Nathan: good job of that. so anyway, those were the main points, but and there's some other other things that sort of f you know fell on the heels of that. But anyway, I really like this guy pragmatic approach. I think this is how AI gets adopted in enterprise, the right way, versus all the hype and the you know, blustery stuff that you read on LinkedIn and X and wherever. you know, sub stack w everywhere else. So anyway, I didn't know it w what do Jeff Breunsbach: Yeah. There's Jay Nathan: you think about that? Jeff Breunsbach: there's two things that stand out. First of all, it's cool. I want to go listen to this now. I think it'll be a good good listen. there's two things that stand out to me. One is I continue maybe to be amazed at how little companies think about change management around these types of things, right? They like, I don't know, talk about it. Probably often they think about it with their customers, but then all of a sudden they turn around and say, Hey, we got you guys a bunch of Chat GPT licenses or a bunch of cloud licenses, like go use them. And then, you know, again, everyone's sitting there saying, Okay, what about shared skills? What about, I don't know, like, you know, how do I do this? Like, what's happened? And so I just think about like how that, like, to your point, like the idea of like the CTO. First of all, I think what's what's also cool about that is, I don't know, it even though there's 700 people, you think about CTO, right? Like he's probably got a stature in the business. Okay, he's got a chief title. Who's approaching this guy about this? And I imagine he's had to, like you said, almost go down to the individual contributors to be like, hey, like, how do I make myself more accessible to you guys? How do we make this stuff more visible? How do we bring this stuff, you know, down here to you so that you can use it? And I so I think it's interesting that he's being tasked with doing that, right? Versus, I don't know, some I some person on the IT team probably could have been charged with, hey, go figure out how we roll out AI across the business versus hey, CTO, like you're in charge of this. but So I think like the the interesting part to me is around the like idea of training and change management and like how they've tried to prepare people for that. And then I think the the second thing that like stands out to me too is just the the prerequisites. I think like to your point, like I think right now there's just too much hype, right? Connect your data, it'll work. Connect just let it access to let it have access to this, it'll work, it'll, it'll do it. And you just you continue to realize that it's just like not. Like you said, like Jay Nathan: Mm-mm. Jeff Breunsbach: you need this ontology, you need this concept of like where's the data store? What are these systems of record? Right. Like you need these things to exist and you need them to be in an architecture or like in a way that is accessible to the right things. We have permissioning to the right things. And so like I like, yeah, the th the prerequisites also stands out to me. but yeah, this is cool. I'm gonna go listen to this this week. Jay Nathan: Awesome. Yeah, let me know what you think. And w we're gonna do something with this guy. He he responded to me on on LinkedIn. So yeah, we'll pull him in and see if we can have a conversation for sure. Jeff Breunsbach: cool. Yeah. Let's talk to him. That'd be fun. all right, I gotta jump. All right. See you soon. You too. Jay Nathan: Cool. All right, dude. Have a good day. See you soon. Bye.