Jeff Breunsbach: All right, welcome back to another episode of ChiefCustomerOfficer.io. This will be coming out on Thursday, May 7th. Jay, how's it going?

Jay Nathan: good. got my morning has been very eventful. ⁓ So story number one, I go swimming, you know, I swim, right? I like three or four times a week. This morning, I was walking, I've finished swim practice successfully, walking toward the locker room to go take a shower. And you see this right here on my head. I walked right into a pole just like full on looking to the left boom.

Jeff Breunsbach: Now I do, I didn't notice it yet.

Jay Nathan: Walked right into a pole. So I've got this big goose egg on my head. I hope I don't have a concussion. If you hear me start to slur my words, let me know. So that was event number one. Event number two was we get a knock on the door this morning and like before eight o'clock. Like who could that be? Police officer at the door. Yeah, I'm like, ⁓ you know, you don't ever want to see a police officer at your door, right? He's either means you're getting arrested or maybe somebody's been hurt.

Jeff Breunsbach: Okay. Yeah.

Jay Nathan: right? And I'm going to tell you about that. Anyway, she had a smile on her face. So it made me feel better. But my car malfunctioned and it's been calling 911 on its own. So that happened this morning and the police came to my house. anyway, that's I mean, it's 8.45 now. So hopefully things will come down from here. But yeah, that's been my morning. Yeah. How about you? How you doing?

Jeff Breunsbach: Yeah. Wow. What a start. Mine's been less eventful, but I had to run the kids to school this morning. And so that means I'm much more of a like, got to get out the door at a certain time, especially because I know we record these on Tuesdays. And so, and I want to get some work done for this. So, know, 715, we're out the door. And of course it's like, you know, we're changing everything in the world to get to 715, you know, and really it's not anything that's different. We're just leaving like 15 minutes earlier, but my my son with the you know the world's changed so that was my event this morning but uh...

Jay Nathan: That's a harder. That's a lot harder than my morning for sure.

Jeff Breunsbach: There's but there is something I think there's something about just getting in the car like after I drop the kids off There's like, you know, it takes me 10-15 minutes to get there and there's like nothing better than like the 10 to 15 minutes with a hot coffee Listening to the podcast that I want like driving home sit in the driveway for a couple minutes, you know There's just like it's just a moment of solace that I've come to appreciate

Jay Nathan: You Absolutely man, you're in the thick of it and you're about to be in the real thick of it. But yeah, we can talk about that later.

Jeff Breunsbach: Yeah. Well, ⁓ I think we've got a number of things to talk about and, ⁓ I'm excited. So I think like the first thing I wanted to talk to you about was this, ⁓ I'll call it AI show and tell that I ran last week. So, ⁓ I think you were busy, couldn't attend, but we, you know, had this idea about trying to get more people showing the things that they're actually building. ⁓ I think this reminds us of many years ago when we were doing customer success and you felt like you heard everybody have these platitudes about like, just go use this or go do this, go build a playbook, go build a customer journey. And nobody knew what those things like tangibly meant, like what's an actual journey, what's an actual playbook. And so I feel like we're at the, yeah.

Jay Nathan: even know what customer success was. We're the customer success versus CX, right? And by the way, question, that's a very good question. I would say like, so sorry to interrupt you, but it's like it was the question was even simpler, right?

Jeff Breunsbach: True. Yeah. And so I think right now we're in this similar, similar stage, right? Like AI is moving so fast that you feel like there's just these platitudes about like, go use AI and there's not a lot of like show and tell behind that. Okay. Like what did you actually build? How'd you do it? So I just, threw up a post about a month ago. I had to unfortunately postpone this event twice. This is like been a long time in the making, but, ⁓

Jay Nathan: Somebody made a comment about that in the email. Did you see that? It was a snarky little comment.

Jeff Breunsbach: Yeah. ⁓ yeah. ⁓ as I, as I, do this out of the goodness of my, ⁓ it's not the goodness of my heart, but yeah, it's, it's, there's selfishly. I wanted it to like, look at what other people are building. So, ⁓ but no, so it was like, you know, I, got, think we got over 500 people to sign up for this thing. We had 150 people live on this call on Friday and, ⁓ we have four presenters and the presenters basically kind of got up on stage. There's no recorders. I didn't allow any recording.

Jay Nathan: I know. Yeah, exactly. You're volunteering your time here, Jeff. Yeah.

Jeff Breunsbach: There's no fathom call recorders, there's nothing in there, there's no transcribers unless somebody did it nefariously. Yeah. Yeah, bootleg.

Jay Nathan: Somebody did it last week. By the way, if you have a bootleg, please share it with us. If you have a bootleg recording, Jay at customer success.io, Jeff at customer success.io.

Jeff Breunsbach: Um, but there's a couple of, so there's a couple of things, there's four presenters and it was, you know, super low key, but show us what you built. And then I would just ask a couple of questions and then I let the audience ask a couple of questions. We kind of move on to the next person and pretty quickly. I mean, we filled up an hour of time by doing this. Um, and so there's a couple of different takeaways. First, I'll tell you about maybe some of the things that I saw and we'll, we'll chat through those. And if you kind of agree with the direction of what some of these people are building, but, um, so first. There was a leader in customer success who built, who is building, I'll kind of call it their own pseudo like CS platform. Like they are, they're in cloud code and they've essentially built their own environment for their CSMs to log into and starting to essentially build out modules for different actions or activities that they need to do. And the main use case or the first one that they really came up with was, ⁓ They never had a great way to pass over product feedback and they're an early stage company. And so they're iterating fast. They're moving quickly. They're talking to customers all the time. And so what they built was a, ⁓ was a, an AI tool that essentially goes through every single call transcript with a customer. And then boils that down into, I'll call it essentially a product feature list or product request list. But that list now has so much more context than I think of the lists that we used to build in the past. It automatically updates ARR because it looks into the CRM and pulls in the ARR of every customer who's asking for this. It logs the actual person who asked for this thing and their email address and ways to communicate with them. It pulls in the context from the exact transcript. So it pulls in the quote that the customer says, and then it lets you click on that and actually go to that part of the call recording so that you can actually see it directly. And then it kind of bubbles all that stuff up into a single line item. And so this transcribe or this, what am I trying to call it? ⁓ this tool that essentially goes through the transcriptions ⁓ is also categorizing these things altogether. it's not sending over multiple product requests, it's kind of deduplifying it into like a single product request. And the last thing I'll mention is ⁓ it will, so then now it can basically send that over to the product team ⁓ or the product team can interact in this environment as well.

Jay Nathan: do the fun.

Jeff Breunsbach: ⁓ And then it also, once that feature is actually built, she built a communication system into this so that they can go actually email those people who asked for those specific features when this thing is built. And that's all done by, you know, their CSMs being able to click buttons and like send that stuff out. So that's the first example. I thought it was pretty cool. And she talked about, you know, essentially this environment they're going to keep building in and just trying to find new tools or features that they, that their CSMs do daily. And it's all, and then the last thing I'll just mention is this. She, she, ⁓ built this in Cloud code. She said it took her two weeks to stand up the first iteration of this thing. And now it's been maybe two months or so it sounded like. it's, she says kind of iterating, but I mean, she said with, prompting and her, you know, it's literally herself. She does not have a rev ops or CS ops person. There's no developer that's helping her. think she said the developers might've helped her set up maybe some of the initial environments. ⁓ So they're all kind of in the same, ⁓ maybe not repose, but they're all in the same kind of environments that, that, ⁓ It kind of plays nice, yeah, so she's like, set this up myself. Took me two weeks. I'm just prompting it and now we're just iterating and I'm asking it to update and things like that. So that's the first use case. What do think?

Jay Nathan: I mean, I think it's really cool. It's people solving their own problems, which struck me about this particular example. And I think it's what we have to see more of here is if you're going to be investing your time into using these tools, have to make it so that more than you can use the tool, which she did, right? Which I really, appreciate about that. And so to do that, there are some concerns that you need to be

Jeff Breunsbach: Yes.

Jay Nathan: aware of like data security, just general privacy, like what, what data can people see versus not now? I think the use case you described is pretty, general, right? But I, one of the things that I've come up against even in building our own internal, we call it our call intelligence system, like our, our, I've talked about it on this podcast before our, our call transcripts in our recordings are the data source for my company. We are consulting firm, right? And so ⁓ But we have hundreds of calls a month and some of those calls are private and confidential and not everybody should be able to just pull them up in quad and ask, you know, what's going on, you know, in private conversations. So we had to train an agent to actually filter that stuff out and make sure it got filtered out and run evals against it to make sure it got filtered out reliably. So there's all kinds of, you know, data. think that's one of the challenges that maybe keeps teams from walking up to these kinds of problems because when they start to sort of gather all the data together, it's like, well, there's all these edge cases that we need to be careful of, which is true. But it does take real actual time and energy and engineering of a sort to, to go, you know, make these systems private and secure. Not in the actual private data privacy insecurity.

Jeff Breunsbach: Yeah. The other.

Jay Nathan: component, which you can't let this stuff leak out, right? It's got to be deployed on a secure platform. Your IT organization needs to approve that platform, but every IT organization by this point, I think should have like provided that infrastructure to their team to be able to do what she's doing.

Jeff Breunsbach: Yeah. Yeah. And like Google OAuth, you know, they're to log in like, like simple, ⁓ or, ⁓ trying to use the tools that, ⁓ I guess are at her disposal. ⁓ the last thing about this use case I thought was interesting too, is that she built it in a way also that she could check her own like, ⁓ billing against it. Right. So I think people are already concerned about token usage. They're concerned about.

Jay Nathan: Yes.

Jeff Breunsbach: All these two, all these subscriptions, you know, whether I go build it myself, is that actually going to lessen the cost of me going to buy a tool that's already out there. And so think she did it in a way that she could pull up a dashboard and basically show. Here's my subscription fees. Here's my token usage. Basically here's my output for this month. And, ⁓ I think pretty clearly she said she's already, ⁓ looking at, know, that, that amount of what she's spending on, spinning up her own environments is already. ⁓ Way less than if she was buying subscription tools herself. So ⁓ I thought that was pretty interesting kind of first use

Jay Nathan: Well, most of the token usage comes in the build for a lot of these things, right? I mean, aggregating a little bit of, you know, customer feedback into a summary is not a heavy token usage kind of operation, especially if you're just doing it once, you know, but there are some other techniques that go along with that, that, can really, really help like, you know, vector databases, short-term memory databases, like there's all kinds of

Jeff Breunsbach: Yeah.

Jay Nathan: componentry that sits around this stuff that again, like we still need engineers in operations folks to build this stuff out and make it bulletproof and make it scale. But I mean, that's amazing that she's already thinking about those. Cause my, from my understanding, I wasn't on call, but she, she's not a technical person, right?

Jeff Breunsbach: Yeah. No, yeah, she's just said two weeks in Claude code and like people were asking her like, how'd you do it? And she was literally just like, I went to YouTube, I looked up Google stuff, like she's, and I just asked Claude itself, like questions. So she was just, you know, kind of like, Hey, I'm, I'm a beginner starting at scratch and just like, I don't, I'm, I'm putting words in her mouth and I don't mean it like this, but I do think that she is, was okay. Asking like kind of the quote unquote, like dumb question to Claude, right? Like, I don't know how to do this. How do I do it? You know, which

Jay Nathan: Yep. Yep.

Jeff Breunsbach: Which tool do I set up? Where do I go? So like, think she was fine, you know, kind of asking those types of questions and trying to just let it, it's got it, got her where to go.

Jay Nathan: It's amazing. Aaron Levy had a post this past week about he's been posting really interesting stuff about services in this world. And like this is an area we're leaning into like AI service, like agent build kind of services. And but the one thing he said is that every is so true. I mean, you can see it happening already. Every department, instead of having its own software, should have its own ops person who knows how to build like this. Right. In some cases, it

Jeff Breunsbach: Yeah.

Jay Nathan: be the manager in a smaller company. And I think there's real power in that because a manager knows exactly what they need to see and what they need to happen. You put Claude code in the tools or in the hands of ⁓ a manager like that, like amazing things become possible. ⁓ But I think they're like all these ops roles that we've had, they've got to just be, they all have to AI. They all have to first, right?

Jeff Breunsbach: turnkey. Yeah. Yeah. Yeah. And I think like, ⁓ I was gonna say like, they should all become turnkey in like, almost like the like you what you were alluding to earlier. It's like, we should have already made some decisions about environments and things that we can build in in a way that like, like, if we've got the data in the right places, like it should become turnkey for us to basically use that and spin up and like spin up things. Like I actually, ⁓ for a while had been dragging my feet on something. And you showed me your HubSpot dashboard that you guys use internally or sorry, you showed me your Claude dashboard that you built on top of HubSpot data. And I just went, I went and did the same thing. I was waiting for like, I was like, ⁓ I got to go build this in plan hat or HubSpot. got to go build these dashboards and I got to go like drag and drop and do all this stuff. And then I saw yours and I was like, wait, what am I doing? Like I can literally just go do the same thing.

Jay Nathan: Yeah, yeah. I will never build a dashboard in somebody else's tool, probably ever again. I think it's big statement, but like why? Unless it's a prompt driven way to do it in the same way that I can do it with Claude, like I'm not gonna do it. It's not gonna do it.

Jeff Breunsbach: Yeah, it's Yeah. Yeah. ⁓ All right. So the next couple examples. go ahead. You want to say something?

Jay Nathan: I had one more thought like this data layer thing is is really like we shouldn't gloss over that that data layer component is really, really important to all of this. And that is the piece that I think needs a company wide, not a departmental level ops person, but actually somebody company wide because you know, you've got all these systems you might have gone, you might have ⁓ what's the one that

Jeff Breunsbach: Yeah.

Jay Nathan: like that Salesforce just bought, ⁓ Momentum, might have Fathom like we use, getting, centralizing that data, whether you're doing it like the way we're doing it by extracting all of that data and loading it into our own central data lake, or ⁓ connecting it via MCP directly to the agent that you use, those decisions are really, really important, both for functionally how it's gonna work and then also to your earlier point about token usage. ⁓ So like there's some real architecture decisions to be made around all this data. And like that is the foundation for making all this work. And it shouldn't be under invested, I guess is my only point. the, it's good to put some solutions together quickly so that you can quickly show what's possible. But I think we have to move quickly toward, like there's a real data strategy around this stuff too. We need to centralize some resources, get them into the same place because it's just going to be way more effective in the long run to have, you know, structure around that stuff. So anyway, just struck me from what you were saying.

Jeff Breunsbach: Yeah, no, I agree. I mean, I think we've talked about for a long time, kind of the idea of like customer data platform and like the idea of like having all the stuff centralized. And like, I think there's probably a huge advantage to whoever spent time doing that the last couple of years, right? Like you should essentially have a distinct advantage to be able to pull this stuff together. All ⁓ right, the other things that I thought were interesting, somebody shared a reference ⁓

Jay Nathan: Thanks

Jeff Breunsbach: tool. they're, you know, essentially built an environment within cloud code that ⁓ allowed salespeople to the sales team to come and look for a specific reference ⁓ from a customer. you know, if I need somebody who's got specific products that they purchased of us, that they're happy and engaging, that their metrics are looking good, they essentially built an environment where like the salesperson could come in. ⁓ It's right now built in a way that's more like toggle style. ⁓ versus like a prompt, but he in essence was going to say like, I want my salesperson to come in and be able to essentially like write a prompt of like, I need a customer who's doing XYZ with, with ABC and then be able to spit out three or four references. And then that can kick off a workflow with the CSM. That's like, Hey, we should go together to this person and see if there'll be a good reference for us for this new customer that we're trying to get. And so they built that on cloud code. It references like HubSpot and other CRM data and ⁓

Jay Nathan: That's cool.

Jeff Breunsbach: you know, kind pulls in product usage, ⁓ a bunch of other ⁓ NPS and other factors. think support tickets were involved in that too. So they kind of ⁓ mashed this together to allow the salesperson. So now they don't have to go bother the CSM with like, Hey, do you have three customers I could talk to? It's literally like, Hey, I have this customer who I think is perfect fit because of these reasons. Can we go out and outreach to them together was their example.

Jay Nathan: It's interesting. All these examples, they're, sort of soft use cases, right? Which is like, give me some more information about contextually what's going on or help me connect with an account or tell me better understand what the feedback is from the user. Like I want to see us get into more hard use cases where, and more agentic use cases were great. We built an app with these tools, but now like what agents did we build to go act on our behalf? Right. And maybe it's prep. And that is sort of an agent, right? It's like the reference agent. ⁓ It's very narrow, but I think that's okay. But like, how do we get to, maybe you got some other examples that are like, no, like this is a real, this is an agent that's actually doing something every day on our behalf. That's critical, mission critical.

Jeff Breunsbach: Yeah. Yeah, yeah, the ⁓ so the one that I shared, ⁓ I built a workflow for my team. ⁓ Again, I say a simple use case, but I think it goes a long way where ⁓ effectively automatically after every single fathom call recording, I have an agent that goes through and it will scan the transcript and it'll look for action items ⁓ from that call. ⁓ We essentially then go into plan hat and automatically build out tasks. And then in those tasks, I have AI tagging it with certain tags to make sure we know what it is. And it's, you know, not based on the CSMs kind of own judgment. It's based on certain tags that I put in there. It tags them. It tags the due date of that task. If it doesn't, essentially I ask it to like to understand how, how complex is this task? How many business days you think it will? And then it puts in a date that it does. And then drops all that context from the call into the description and links out to the Fathom call recording into the transcripts that you could actually go look at the context yourself if you need to. ⁓ So it does all of that. And then it also builds an email inside of our email planner ⁓ within PlanHat that essentially will, the CSM can go push send on. And so it ⁓ uses AI and goes through the call transcript and says, pull in any details that they talked about at beginning of the call. you know, some, some niceties, Hey, great to talk about your son. Hey, awesome. That the weather was awesome, whatever. And so it kind of builds some, some stuff up top. And then it has like the, the context at the bottom of the, essentially the tasks and everything else. So, ⁓ I think it's super simple, but like, you know, my team now literally doesn't spend any time logging any tasks. ⁓ it, know, they're essentially logged into their platform and it runs automatically on their behalf after every fathom call recording. So it's like one example that I've got right now.

Jay Nathan: That's cool. Yeah. then, yeah, then they can use that list to check items off essentially over time.

Jeff Breunsbach: Yeah. And so like the next, and so like what, what my team and I were talking about too, is like the next, the next iteration or V2 of that is like, could you then for those action items, ⁓ let's say there's internal, like I tagged one internal coordination. Like I need to go coordinate with my product team. Can you actually draft the message and go put it into a specific channel and already do the internal coordination? So like me too, my team's already asking me about is like, okay, can you just go draft this Slack message for me about, you know, what product this and like bring the context in and like, do it. like. Yeah, there's already, I think my team's already like spinning the wheels and like, okay, how do you take it to the next step to have agents working on agents that can go do something like on their behalf while they're on another customer call or something.

Jay Nathan: Yeah, yeah, exactly. ⁓ I have a proposal skill in Claude and I have obviously our call intelligence system and what I have not done yet is built the agents that literally just take the call and say, hey, this sounded like a discovery call. Do you want me to build you a discovery summary and a proposal for this? No, I have to go do that manually, right? ⁓ So that's an issue that I think I can easily solve.

Jeff Breunsbach: Yeah.

Jay Nathan: think what I'll do there, I'm just thinking out loud now, is I could start by solving that with just a scheduled routine in Claude today and have that become my proposal draft agent.

Jeff Breunsbach: ⁓ well, I mean, it's similar to what I did too. I wonder if you could just do it after every fathom call recording. Cause I imagine after every call, do you go through and categorize what type of call that was and put tags against it?

Jay Nathan: Well, I mean, I do for, um, for security purposes, like I talked about earlier, but what I don't do is try to categorize it by what type of call it actually was. Was it a, was it a discovery call? Was it a scoping call? Was it a proposal review call? Was it a consulting call? We could easily do that, but I think you could also do it before you get to that level of sophistication in like the underlying.

Jeff Breunsbach: Yeah. Yeah, yeah.

Jay Nathan: Guts of it. You could just have an agent run in Claude or whatever tool and then go pull all the calls that have happened over the past week and tell me which ones you think were. Discovery calls and you know, but I think. This is where like. You want to sort of unpack your process a little bit and try to say, OK, like what's important to automate here? Like what's what's the repeatable thing that we need to need to do and? ⁓

Jeff Breunsbach: Yeah.

Jay Nathan: And then go start to build an agent to do that repeatable thing. It being generic, like I just, you know.

Jeff Breunsbach: Yeah. So one, ⁓ so I just built a renewal workflow with my team. And again, the V1 is like, is the renewal coming up in 90 days? Yes. ⁓ And then I have Claude go through and I have it start doing a couple of things, but basically like, I want Claude to do a summarization of it. And then I want it to come back and give me some kind of, I'll call it markers, but like, is the customer using the product and happy? have they been engaged with us in the last couple of months, like a couple of those markers and based off those markers, ⁓ it essentially gives my team a list of tasks to basically close that renewal. So we're 90 days out, but the tasks change based on those markers, right? So if customer is not adopting the products, I don't want you to have the same seven tasks that another healthy account does. I want you to figure, okay, now we've actually got to go front run this. We have to go figure out why they're not using the product or like what usage needs to increase or whatever. So that's like one element to it. ⁓ then, but then I guess like V2 that I'm already thinking about is like, I want them to schedule a specific renewal call with the customer, talk through the renewal, talk through how they've been the last year, you what's important to them, whatever else. ⁓ And then V2, like what we've already talked about with my team is like from that renewal call, the Fathom call transcript is going to come in. Can I tag it renewal? You know, renewal discovery. And then based off that renewal discovery, Can I take our template proposal that we have? Can I take their current contract? And can I take this renewal call and can I basically mash those together so that it creates a Google doc link to a updated proposal with pricing and stuff built in for my team? And then like at least the first step is done, right? Like I don't have to go build a proposal. I can basically just kind of have a, my team can have a V1 rather than starting at zero or. I think the age old question that drives me nuts right now is like, where does that template live? How can I go pull that template for the proposal? so then can I have that agent as like the V2 basically come in and say, okay, here's your Google Doc link and actually attach it to the deal record. ⁓ And that's the proposal link and my team can then go present that to the customer. We can adjust it how we need to, but essentially it tries to take out some of that early ⁓ thing that they would have to do.

Jay Nathan: Yeah. think this is all iterative, right? You build the renewal agent and you have to, this is back to the idea. You need to have somebody on your team, whether it's in a small team, the team leader or an ops person who's continually sort of refining and it becomes a product internally, just like anything else, right? I mean, this is why I think, you know, yes,

Jeff Breunsbach: Yeah, yeah.

Jay Nathan: these agents will replace a lot of the labor that we do manually, but you still need to have somebody to continually build and refine the agents. Like I've even found in some of the agents that I've built over the past three months, I'm like, or six months even, I'm like, it doesn't really work the way I want it to anymore. Like now I'm smarter than I was six months ago. So I need to iterate this. Now I've got to spend some time on that. ⁓ And I'm trying to run a whole company, right? So. I wish I had somebody who was sort of helping me tune these things as we go and paying attention to these processes as much as I am, but we're small today. But, ⁓ I think that's why you need a partner. Like you might need partners to do, or people who are in your organization to do that kind of work, whether it's your own people or third party.

Jeff Breunsbach: Yeah. Yeah. Yeah. Um, actually we can come back to that in a second too, cause I'm curious how you are capturing some of this stuff to work on. Uh, and then let's see a couple other, uh, to come out of that, that show and tell last week, a couple other call or a couple other use cases where some MD, um, kind of examples, just some skills that, know, your team could be using in cloud coworker, cloud, uh, dot AI. Like one of them was a, um, I built a product marketing skill. And so you can essentially give it any sort of link or other internal documentation. And what it'll do is it runs through a set checklist of like, um, five steps and the output essentially becomes a enablement guide for the CS team and customer facing assets. And so the customer facing assets become slides. becomes a PDF, it becomes whatever. like trying to, um, uh, not rely on our marketing teams or product marketing to always have to like, you know, hand feed us stuff. Like I could actually go, like I showed the example, like I went to our API documentation. I pulled a link and said, ⁓ here's, here's a new thing that we released via the API, dropped it in and it spit out, ⁓ you know, more of a, ⁓ more of a non-technical version of a customer facing asset that says, Hey, if you're trying to accomplish this, we have something that you could do. And here's how your engineering team can basically go do it. So ⁓ some skills like that. think some other people's shared some other skills around coaching for CSMs. ⁓ you know, coaching like a coaching skill that they can coach their CSMs with and some others. So there's like a couple MD style things that people were showcasing.

Jay Nathan: What about skills that CSMs can use to coach themselves? Back on this call based on I am.

Jeff Breunsbach: Yeah, could do that.

Jay Nathan: the other day, I was reviewing an interview that somebody did. I was reviewing a call and I was like starting to make notes and I was like, you know what? No, I just need to like provide coaching directly in in quad based on this transcript. And then turn that into a markdown file and call that the interview coaching skill. And that people can, you know, say, wonder what Jay would think about this interview or how like what feedback he would give me on that. Be in the room, right?

Jeff Breunsbach: Yeah. Yeah. Yeah. That's cool.

Jay Nathan: But you gotta spend the time. You gotta spend the time. This is all enablement, right? We're enabling people through documentation now, essentially.

Jeff Breunsbach: Well, the, and I think the interesting thing, and maybe you might know more about this than I do, but, ⁓ the interesting thing that we kind of talked about on the show and tell too, was like, ⁓ once you move to an enterprise wide org, it unlocks some capabilities, especially, think for cloud AI and cloud coworker, you can essentially share the same skills. Cause right now, like we talked about some of limitations of like, ⁓ I have to download this skill, send it to my teammate to basically upload and use. But, ⁓ if I remember correctly, like,

Jay Nathan: Yeah. Yes.

Jeff Breunsbach: enterprise wide org, you basically have a shared level of skills. And so that too means that you can refine that over time and say, okay, like now we're all using the most up to date skill and have, you can probably actually prescribe a little bit of a process and almost ownership to some of your team members. Like, Hey, you're owning this skill. Like, know, you, you've got to make sure it's up to date. And if we change anything like

Jay Nathan: Yes, yeah. Yeah. Yeah. Part of your job, part of your job is to maintain these skills for the team. And if they get out of date, like that's on you. If you're the manager and you provide coaching skills to your team and you expect them to use them, like, how about this? Like I'm the manager. I have a set of coaching skills that I've built for you. My expectation is you're going to come to our one-on-one if we even do those anymore. But if you have a monthly one-on-one or whatever, and you bring me the feedback that you got from that coaching skill, let's talk about it.

Jeff Breunsbach: Yeah.

Jay Nathan: What did you learn? Right? I mean, what a constructive way to work with the team, right? ⁓ I will say that the thing that so yes, by the way, like what you just said sort of speaks to where everybody is on this on the continuum here, because you should absolutely be using the team version of Claude, right? You have to put skills in at the team level. You can't be passing around markdown files. ⁓

Jeff Breunsbach: Yeah, that's a good, I'm gonna take that actually. That's a good one.

Jay Nathan: We are, I talked a little bit about the Balboa OS. That's like our operating system. Like we've begun to document tons of information around our company, our go-to-market strategy, our marketing strategy, our service delivery model, our services packages in a set of markdown files. Like our company is becoming code. And then that is going to be distributed automatically to everybody's machine via, we check it into GitHub.

Jeff Breunsbach: Yeah.

Jay Nathan: GitHub automatically sends it out to, ⁓ we happen to use Microsoft, which I load every day, but we do. And so we automatically send that out to OneDrive and then OneDrive automatically distributes it back down to everybody's machine in the company. So everybody has the same set of Balboa OS files. And if we want to change something at the company level, maybe it's our handbook, right? Our policies, our vacation days, the people on our team, like the roster. the people at our partners that we work with, like it's all in there and it's distributed out. So if you have a question, you just ask Claude cowork. You make sure Claude cowork is attached to that folder. When you start a new task in cowork or a chat and, then it's all right there for you and it's all up to date. And a lot of our skills are embedded in that as well. So you need to have a distribution strategy, whether it's through Claude or through some.

Jeff Breunsbach: That's cool. Yeah. Yeah That's cool.

Jay Nathan: know, file sharing strategy of getting all these tools on people's computers.

Jeff Breunsbach: We should talk about the mechanics of that maybe next week, because I think people will find that super interesting. Like your idea of pushing this stuff to computers and like how everyone's working out the same. I think it'll be a good topic. Okay, so.

Jay Nathan: the other thing ⁓ was thinking about real quick on that is when you build these skills, like one of the things I've found recently, because we've been on Claude Claude just released Opus 4.7. It went backwards and they've also been throttling their like they've been defaulting to simpler models and sort of throttling the thinking. ⁓ So when you build these skills, evals are really important, meaning an eval is basically like a unit test for a skill.

Jeff Breunsbach: Mm. Yep.

Jay Nathan: It's like, hey, here's the success criteria for whether you did this right or not. And I think it's important, especially as these models, you have to keep in mind the models and the services underneath you are changing. so what worked one way when you built it, this is a probabilistic system, right? It's not deterministic. It's not going to work the same exact way every single time. So you have to build with that in mind and make sure that your skill that you built three months ago is still working the same way. That's a real thing. So just something I learned in past couple of weeks.

Jeff Breunsbach: Yeah. Well, that's like, and I think that's a big, I guess, call for like the humans in the loop, right? Like you want to still put, like if something, I still think you want to put something in front of a person, right? Because, but, but the more that you can take out of them having to go kind of do the manual work to get the information, to pull that stuff in, right? Like that's what you're trying to optimize for is like, okay, if I can just put this in front of Jeff and Jeff can then determine.

Jay Nathan: ⁓ yeah.

Jeff Breunsbach: Is it the most accurate up to date relevant information and then like then can press the button like that's I think that's where the wins come. All right, so I think like the big thing that came out of this for me was ⁓ the idea and sense of community. know, we years ago started posting on LinkedIn. We had our own business at the time. We built a community called Gain Grow Retain and ⁓ you know, since it's we've kind of exited that business and not. not been involved in that community. still think we're involved in many different, I'll call them like sub communities. I think you and I are kind of in a bunch of different spots, LinkedIn and others. But I think what came out of that call for me, like there's two things that came out. First of all, got, so 150 people showed up. I think now I'm registering a 50 % response rate on my NPS. So 70-ish people have responded to my NPS. And I don't really care about the number. I'm not gonna tout the number, even though it's 4.9, pretty good. 4.9 out of five. More so what came out of that was the comments, which was like, you know, it's valuable to see people building. I want more of this where I can basically see other people's environments. I can see what people are doing. ⁓ and so it kind of left this idea in my mind of like the void that I think still exists around communities and whatnot. And, ⁓ I think it also just reinvigorated like my passion for communities, you know, like I, I enjoy meeting new people. I'm, know, constant tinker and learner and I just like kind of thrive in those environments. It just to me is like super fun to be able to be curious and ask questions. And so I came with you, I came to you with like a hair brained idea, which is basically like, Hey, we have a podcast and we do this newsletter. I really think we should be doing community. And so we're to launch uncommon, which is going to be our community around AI forward customer leaders. And the whole concept is how can we bring people together? who are on the cutting edge of using AI to help their teams be better and help their customers get better outcomes. ⁓ And so to me, I'm excited about this because what we're also talking about doing is building our own custom environment for this, We're gonna use Cloud Code and almost like dog food this stuff ourselves, which ⁓ I've been mainly using Cloud AI and Cloud Cowork. And this weekend I spun up a Cloud Code instance and like, was like pretty blown away of like, ⁓ my gosh, I should have been using cloud code way earlier. Like I, you know, like you said, like there's nice wrappers on cloud AI and cloud cowork and like things can get done and I can connect the tools, but there's something very different than about moving to cloud code and being able to essentially like custom build apps in environments and things that like, you know, ⁓ are pretty in my mind, I guess, like eyeopening of like, okay, like, wow, like they're almost like the possibilities are endless, which I think can be, you know, good and bad, but, ⁓

Jay Nathan: Yeah.

Jeff Breunsbach: But yeah, so we're excited. So we're going to, we're going to launch this community uncommon. ⁓ It's going to have a ton of features or like benefits around it. But really the first two that we're going to market with is going to be a legit community so that you guys can engage with each other. That we can, you know, put ideas out there. We can essentially, you know, have a way for us to communicate. And then the second thing is just these show and tells and trying to build on some of these ideas of like. kind of real live events, how do we kind of bring the community together more often? And so those are like the two big things that we're going to come out with. But yeah, I'm excited. I think it's like super invigorating for me. like, you know, think for a while it was, ⁓ I am working in, you know, worked for the last couple of years and feeling like, ⁓ man, I'm kind of doing this stuff myself and I don't know who to talk to. And like, I'd have some one-off conversations, but you never really felt like maybe like those were enough. So anyways, I'm excited. I think this is gonna be really cool.

Jay Nathan: Yeah, it's going to be great, man. ⁓ We. I think Jeff and I were most invigorated, probably probably most invigorating point in our career. There was an inflection point during Covid where it was clear that there was a need for community and, you know, is a little bit. A little bit induced by what was happening at the time, but this is a whole. This is a whole different thing. we're just, it doesn't seem like anyone has really stepped into sort of take the lead on this in the, in the customer world. And it goes so much further beyond SAS too, in my mind, which is the exciting part. mean, part of what we've found Jeff right in GGR was like, we have a lot of great friends that we made that were setting up customer success in industrial services companies. And Like they were just adopting these mindsets and strategies and they looked to us as like the facilitators to help them figure out where to go to get information to do it. We have never considered ourselves to be the people that have all the answers, right? And I think that ⁓ that will come through and what we're going to do with uncommon, which is we're the DJ, not the talent here. And our whole goal has always been to take other people and put them in lights. And we've made such great friends. mean many of I mean these we talk to these people all the time, right? And they literally are our friends now. ⁓ But they're doing really cool stuff, ⁓ so I'm excited to take whatever little platform we have ⁓ use it to amplify other people ⁓ ⁓ know, make something uncommon and special, so it's going to be cool. ⁓

Jeff Breunsbach: Yeah, chiefcustomerofficer.io slash uncommon. If you want to sign up and show some interest right now, we should be launching a few things here in the next couple of days.

Jay Nathan: Yep, and it all hinges around. We always have sort of a mantra with anything we did, whether it was a live event in person or a virtual event. Just make one connection that you didn't have before to somebody else who's in a similar place that you are and get one idea that you can go implement tomorrow, right? So it's like action oriented and relationship oriented in it. And I think Jeff, luckily you and I just I think. where our personalities are just geared toward that relationships and action. So it sort of works out well. ⁓ but I'm excited to see how this thing can, it can unfold. One of the things we've, we also did with GGR that was just, don't know. It was just the way it unfolded is we didn't try to plan the whole thing out on day one. When we launched GGR, said, you know what? There needs to be a community here.

Jeff Breunsbach: Yeah.

Jay Nathan: We need to do office hours. saw who was it? We saw doing that. It was it. It was the guy from Gainsight, the marketing, the old marketing guy. What was his name? Yeah. Anthony Kanata. He was doing a marketing office hours and we were like, Jeff, we should do that. We did it. We had 30 people on the first call. Then we had a hundred people on the second call. Then we had 200 people on the third call. was like crazy. And I think, um, and then we built all kinds of other things around it. And so

Jeff Breunsbach: Anthony Kanata?

Jay Nathan: We'll build this in public too, right? Jeff, mean, I think that's the cool part for us is we can, we can build this and, you know, ⁓ see what, see what blooms on top of it as we, as we do.

Jeff Breunsbach: Yeah. No, I'm extremely excited. And I think the last thing that I would say is like uncommon and kind of the reasoning behind, I guess the name or what we're thinking about is that like, in order to get ahead right now, you've got to be the one who's making uncommon decisions. And like, that's the only way that you're going to get ahead is that you can't kind of go with the norms. You can't kind of go with what's been happening. Like you've got to figure out, okay, what's the bold thing that I can go do that's going to give me an advantage. Example, you know, this woman built her own kind of CS tool that, you know, she's using and building on top of her team. Like that's, that's probably not an easy decision for her to go tell her CEO, Hey, I'm not going to go buy something that's off the shelf. I'm going to go build it myself. Like that's an uncommon decision. And like, but that's going to give her an advantage and she's already been finding it and seeing it. And now she's building on top of it. And so, ⁓ you know, uncommon is basically our, our pitch about you becoming a better leader is making uncommon decisions and building your business of the future.

Jay Nathan: I had somebody yesterday. I was talking to a good friend from the community, GGR back in the day, still a good friend who just inherited ⁓ customer success organization. He was an ops. ⁓ he's running the whole team. By the way, there's a lot of that going on too. Ops leaders are ⁓ and I think both sales and customer success, but he was like, I am not going to just go do this. ⁓ can't just go do this the same way. Like that doesn't make any sense. Right. And he's

Jeff Breunsbach: Yeah.

Jay Nathan: Absolutely right. It doesn't make sense to go do it the old way. So this is, you know, build the infrastructure first, build the scale first layer in humans where it really, really matters. By the way, humans are still a moat. There's a moat, right? You can automate anything you want, but human interaction cannot be replaced still. ⁓ So anyway, I love uncommon Jeff. credit goes to you for coming up with the concept. guess maybe you and Claude a little bit of a brainstorming session.

Jeff Breunsbach: Yeah, we did. We went back and forth a little bit. ⁓ yeah, I'm excited. So we'll keep sharing more details about it on here. We're going to be building stuff in public, like you said. so anybody who wants to some interest, ⁓ chiefcustomeropster.io slash uncommon. And ⁓ I think we'll wrap it up here for this episode. next week, I think we should try and talk through a little bit of ⁓ the concept about how you're pushing some of the files to your team. I think people would find that super interesting to make sure we do that. OK, cool.

Jay Nathan: Yeah.

Jeff Breunsbach: We'll see you next week.

Jay Nathan: Cool, all right.