Jeff Breunsbach: episode of the Chief Customer Officer Podcast. This is coming out to you on Thursday, October twenty second. Jay, how's it going? Jay Nathan: It's going great. How are you? It's been a been a busy couple of weeks here on my end. Jeff Breunsbach: Yeah, same. I feel like we're in the heat of also what's about to be about to be holiday season. And so I'm I'm you know, excited for Q four but nervously like, gosh, like we have a lot to do before I feel like the proverbial like everyone eats, you know, eats around Thanksgiving and then it's gonna be New Year's and all of a sudden you're like, wow, it's already twenty twenty seven. So Jay Nathan: Yeah. Does your fiscal year end in December or January? Okay. One Jeff Breunsbach: December thirty first. Yeah. Jay Nathan: of the decisions we made with my new company is January thirty first. So I will never Jeff Breunsbach: Nice. Jay Nathan: go I will never go back to a December thirty first year in close. Be I hope. I hope. Yeah. Yeah. Jeff Breunsbach: That's a that's a good thing, I'm sure. well, we were just talking. I think one thing I just showed you was like our our platform we're building for my customer success team. And you know, the I guess the quick update right now is like we have a platform and a product that my team is using, which is like exciting. It's in people's hands. Also very daunting, by the way. I was like super self-conscious about releasing the product and also showing our Jay Nathan: interesting. Jeff Breunsbach: Like doing our engineering team and stuff, just because I'm not, I'm not an engineer by trade. I I got this I got, the imposter syndrome of is this is this going to be good enough? Are they going to not like laugh at it, but are they going to take this seriously? Or are they gonna look at this as as almost like, I guess beneath them or something? But Yeah, yeah. But like I I guess like at the end of the day, I finally was like, you what? Like This is the best thing for us as a business. we need this. I need and no one else is gonna go do it. So this is what we have. Jay Nathan: Yep. Let me ask you this. sorry, go ahead, finish your thought. Jeff Breunsbach: and I was just gonna say, and like at the end of the day, you need to get this l like I I kept kind of fighting this deadline like, okay, get I I need I need to perfect this, I need to manicure this. And it just go like I got this revelation because I looked over at my bookshelf and you know, zero to one and like all these basically all these books that are basically like, dude, you need to ship something and get it in people's hands and then you iterate. And it's like the whole ethos. And finally I just got over it and did it. And you know what? Feedback's been people have found stuff that are that's broken and it's like, you know what? Like we needed to we needed to do that. You know, like that's something that we needed to like actually do. So and then our engineering team has been great and helpful and in trying to, not necessarily even poke holes, but just make sure, they helped me set up some of the controls. Hey, it's it's only accessible via VPN and we've got some of these security and controls in place to make sure that that's going well. And then but then I've showed it the the best thing I've done is I've showed it just to people across the business. And everyone now is starting to see maybe the power in like pulling together context on customers into like singular places, right? And that's been the exciting part is people have had these ideas of like what we could add in there, you know. Can we add usage data? Can we add other can we add configuration details that come straight from our database, right? Like so that we don't have to keep that updated. so anyways, built my platform, it's been fun. what's your question? Jay Nathan: Super powerful. Well, my question for you is do you get self-conscious when you show your executive team a spreadsheet? Today? An Excel spreadsheet? Or sheets? Okay. So when Jeff Breunsbach: not really. Not really. No. I don't think so. Jay Nathan: you just this is the new version of a spreadsheet, right? I think Excel is the most prolific application software application development tool on the planet, up until Jeff Breunsbach: Yeah. Jay Nathan: AI, right? And So what you're doing now is going to be that is mainstream. That will be mainstream. I think everybody needs to be building their own stuff. I'm I'm more convinced on this now than I ever have been. Jeff Breunsbach: I would agree. And also I'll give you another quick, quick use case or example. we have a part of our product that we know we need to build out. And it's like a it's something that like works today and is usable, but there's a number of features that our customers want and need, but we've already talked about how we're not prioritize like our product engineering teams are not prioritizing it for Q4. And so what I said is I said, Okay, well, it looks like two of our customers are gonna need this in Q4. Our product and engineering teams are not prioritizing it. And that's fine. That's a business decision we made. We're all comfortable with that. But Jay Nathan: Yep. Yep. Jeff Breunsbach: is there actually an interim solution that I can build that looks very closely to what they need to build into the product and and basically almost gives them early users, right? Like, and so I with my CSM who works on these two accounts, I'm basically building this UI for her that is going to do the feature set that we need or our customers will need. And she's just going to handle it for two to three to four locations. And I showed it to our product guy and he was like, dang, that's pretty cool. And like this is this is going to be helpful because now our CSM can use it with the customer. And like we can basically say, Hey, this is like a kind of an alpha or a beta, right? Hey, we're just we're we're trying to figure out what's the workflow, what's the UI look like, what do you need? And we're actually just gonna do that in real time together, like this. And we're still storing the information. We can make sure it's captured and whatever else. But now, do we have a leg up come Q one where we say, hey. We've basically already scoped this thing. We've actually already built the UI. And we've actually already we've taken this almost to the to the furthest point where now it just needs our engineering team to to codify it and say, okay, this is repeatable for hundreds of customers. This is durable. This is the right thing to do. Here's the database structures we need. Here's the information store. Blah, blah, blah. Here's the API calls we need. Boom. That's how we get it into the product. But I think that like I'm almost starting to envision how customer success teams, and I have a solutions engineer on my team, which I think is also a cheat code, right? But solutions engineers and in customer success, it's like, cool, can I actually be building again? I'm not saying we're going to take over the entire roadmap, right? But there's probably three to four to five things each quarter that a customer comes to you and says, Hey, I'd really love this. And you're like, Well, that's never going to get prioritized. But I could probably use our solutions engineer myself and our CSM, and we can probably get us. seventy to eighty percent of the way there and then give it off to the product engineering team when they have the capacity to do it. Jay Nathan: This is the whole idea behind Ford deployed engineering, Jeff Breunsbach: Yeah. Jay Nathan: Like that's it. What you just described is it. So one of the other so I'm I'm writing down like my current laws of software, right? First of all, Jeff Breunsbach: I like that. Yeah. Jay Nathan: software is free. If you know what you're doing, you should get your own software for free, right? Just like you're doing with your CS platform. And you're hooking into some core system of record kind of stuff to build your platform. That's cool. but software is free now. It is free by all intents and purposes, it's free. Okay. So the other thing is that I think, and you're hitting on it right here. I think all software companies should become Ford Deployed Engineering Orgs now. I I wasn't convinced on that at first, but but if you sort of reason it backward, right? If you think that package software is going away over time, right. In favor of standardized or solutions that are custom to my organization, right? your customer success platform that you just showed me is connected to your data sets, your particular systems. You don't have to ask, does it integrate with this or that and have some rigid integration? It is sp it's got your language in it that's specific to your customers. It's very, very specific to you, right? Now it's got some fundamental components that are s standard, right? Some maybe some data models around how you would think about accounts and contacts and customers, right? That you get a lot of that from HubSpot. That's your sort of your structure. But Every company has the opportunity to tailor their stuff to themselves now, right? Their needs to their specific needs. Their success platform should be theirs. So why shouldn't every software company that lives in the old SaaS world become a basically a custom dev shop for their customers? They could still bring the IP back into their standard template, right? And have shared services and agents that are consistent. And aggregate the best practices across the industry because that's the value now. Jeff Breunsbach: Yeah. Jay Nathan: But the software itself is worth zero as a standard Jeff Breunsbach: Well Jay Nathan: package. Jeff Breunsbach: and I think to your point, right, this is this has maybe always been the case, right? even when we were deploying software to businesses, Jay Nathan: Hundred percent. Jeff Breunsbach: right? When we were deploying software to businesses, the the number one challenge was getting those businesses to adopt my software, embed it into workflows that they already have, make sure it drove actual bit like business outcomes, right? Those things existed even when we had software. But I think like the the point you're making, right, is that That was always at arm's length away. And that was always commoditized to the, okay, how can I sell this to the most amount of people? Right. And now it's truly, hey, not only can we get you 80% of what every other what we're doing for every every other business, I actually can go the last 20% and I can make this so tailored to you that and you can actually see the outcomes of this, right? Like you can actually see the benefit of like what we're doing here. like I just thought about, you know, the the way that you could start tying this stuff. altogether inside of the platform as well, right? You like the the tracking or the ability to have a through line throughout this whole thing becomes much clearer and easier too. in all this stuff, in my opinion. Jay Nathan: Yep. okay. The last thing I'll okay. No, no, go ahead. Jeff Breunsbach: One other go ahead. one other thing that I keep that I have thought about too. pretty and I don't know, I don't have the answer quite yet. But I think one of the things that I I keep struggling with, and I'm I'm curious if you've seen this, is that you almost need to it's almost like you need to change the entire mindset of your of your team to basically be, could I use AI first, like to do this, right? And there there still is this disconnect, I think, of like I'm doing this, I'm pushing this button, I'm do it. There's still this disconnect of what is possible. And I think it goes back to this idea that and even when you again before the advent of AI, I think this still existed, right? Which was I didn't know all the possibilities of Salesforce. I just knew what I've explored, right? And it's really hard for me to know, it could actually you could actually do a CPQ that does the X, Y, and Z and has automations and does all this stuff, but unless I saw it with my re my own eyes or somebody from their business came to me and said, Hey, you can do this for your business, blah, blah, blah. But there is there still is this disconnect for me of it's almost now the possibilities are almost limitless and everything Jay Nathan: Endless. Jeff Breunsbach: that like you you kind of get into this thing of like, okay, d am I gonna build something that actually isn't that useful? But at the same time, like again, I I think like there needs to be this bottoms up culture of just like Every day that you do something with AI, just share it in a channel. And it could be the smallest thing in the world or it could be the biggest thing. But I'm right now toying with the idea of creating a channel for our company that's just called How I AI. And the sole purpose of it is I want screenshots, I want skill files, I want like I just need real examples so that people can see, okay, hey, I saw this support person. just do this X, Y, and Z. I wonder if there's a similar c use case for me in success. Or hey, I just saw, hey, I just saw our CEO. you know, he just went and transformed our product by doing X, Y, and Z himself. Like I there, but to me, I I don't know, like there just needs to be almost like this. I I keep thinking of like the new social media activity feed should just be a feed of like what people are using with or building with AI, like the real examples. Like if I had a feed of that, like to me that would just be like a game changer, I feel like, for people to to get on board with it. Jay Nathan: I'm toying with just giving everybody in my team clawed code. They already have access to it, but teaching them how to use it and then giving them a Jeff Breunsbach: yeah. Jay Nathan: place to deploy things because you're right, like a lot of this is iteratively uncovering layers for yourself. I I was actually thinking though, part part of what our ops teams need to become is this is their job first and foremost. there's the strategy piece of ops, right? Which is, you know, how are we measuring the results and the team and that kind of thing. But you have to have the data in place first. You have to have the tools and the measurements in place first. So ops should be building this way, AI first, you know, building Jeff Breunsbach: Yeah. Jay Nathan: the tooling you need. And then hooking in all the components that you need. Like you, I think you and I play that role for our teams because they're relatively small teams today. But Jeff Breunsbach: Yeah. Jay Nathan: that You need a you need someone running ahead of you doing this constantly. Jeff Breunsbach: Yes. The reason that I brought this up too is I saw a tweet from is it Boris who is like the Claude, the head of or like one of one of the lead engineers who's come out there, Boris Channel Yeah. Sometimes Jay Nathan: Yeah. Boris Chang Changgy tri Changy turn off. Jeff Breunsbach: he came out, I saw a tweet that he had over the weekend and he put out some he put out some examples, like a s a video of his screen and how he was interacting with Claude and people came in the comments and were basically like, There's no way that your prompts are that simple. And then that's what you get and you you get Claude to use. And he's like, he he came back firing. He's like, let me show you all the screenshots. this is literally how I talk to Claude. And again, I think in my mind I had this idea of engineers, they must be they they must be supercharging this thing. They've got f software factories. They've got skill files up the wazoo. They've got all this stuff. And literally I like watch him use it and he's he's like He literally put in a two-sentence prompt that was like, Hey, I need you to go check on this, and I need you to update me on X, Y, and Z. And I need you go do this. And he's like, This is how I use it. this is this is the way it's meant to be used. you should. Jay Nathan: But is that because he's got the right tools connected to it, the right context already? Yeah. Jeff Breunsbach: Yeah, like the there's the I think it's the yeah, I think I think it's the he he's clearly got the skill files, he's clearly got the integrations, he's got the artifacts, he's got the context. That's probably all baked in. But I do but I think like, I don't know, in my mind I I have I think I guess come back around to this idea of just like all you need to do is just chat with this thing and like it will continue to do what you need it to. You don't need to Sometimes I just get lost in this idea of like, okay, I have to optimize this thing in order to get the right outcome the first time. So I'm not wasting tokens. So I'm not doing X, Y, and Z. And it's like, just literally go prompt it and use plain language, use what you need, tell it what you want. so I'm like background on that idea too. Jay Nathan: Yeah, that's incredible. it it's funny because there is you don't wanna anthropomorphize, but it is l a little bit like people, right? If you give it too much instruction, then Jeff Breunsbach: Yeah. Jay Nathan: it it sort of okay, it it it sort of takes it too literally what you said, right? Like I find this when when I when I ask, you know, my team to do anything, right? The less instruction I give better the outcome usually is, right? As long as the intent is clear of what we need to do, the outcome is better than I anticipated if I'm lighter on the details, which is a really interesting and difficult leadership lesson. But, you know, you can practice this with AI, is the way I think about it, right? Jeff Breunsbach: Yeah. It's like yeah, you've got a little bit of a runway with AI, compared to your teams probably. Jay Nathan: Yeah. So the ops team is the AI team in my mind. It's gotta be. And you know, Jeff Breunsbach: Yeah. Jay Nathan: there's there's more than just building in tools to ops. There's there is strategy and measurement and some of those other things as well. But Jeff Breunsbach: Yeah. Jay Nathan: the tools underneath that are how you get to the measurement, how you get to the strategy piece of operations. And so, you know, in a large enough operations team, you should have multiple people who are doing advanced things who are optimizing, right? I think we Jeff Breunsbach: Yeah. Jay Nathan: still want the teams to build. We still want the CSMs to become FDEs if they can, right? To to build tooling and things that we need. Start the idea. But then at some point when you need to optimize that, it's it's we need to give it to somebody who is a pro. Our team is doing that now. No, you know, plug. But like you need to give it to a pro who can actually build the plumbing and make sure that the right data sets are in the right place, the right you know. data has been pre-processed so that it's token efficient, all that kind of stuff. That comes later though, right? Like Jeff Breunsbach: Yeah. Jay Nathan: back to your point, you do the prototyping first, you solve the problem first, and then you reverse engineer it and say, okay, how can we really now scale this at a cost effective clip? Jeff Breunsbach: Yeah. the l last thing I'll mention real quick and then I'm curious, you know, if you've got some stuff over the last week that's been curious to you is I've got a ship it list now with my team. And the whole idea of the ship it list is if you find yourself doing something more than once in a week, it should go on the ship it list. Jay Nathan: I like that. Jeff Breunsbach: And the whole idea is then I have a tiger team that I'm calling, where I bring two of my CSMs, myself and our solutions engineer to a call on Fridays. And the whole idea is to see if we can ship something in an hour, together. And it's in again, these are little things like this is little things like, hey, we have to submit a pricing request to one of our partners for a pricing of a marker. And right now, that's a very manual process. We have to put it into a spreadsheet and email somebody about it. And that's just right now, there's no way to fix that. So, but can I automate all those parts or can I make it as easy as possible for my my team to press a button? and do that stuff and and automate the back end so they don't have to do that part. those are the types of things that we're trying to find. And to me, like that's it's been another like, I don't know, this whole idea of just like can you how do we move quickly, ship stuff, get excited about these things. Like this is like almost like this rhythm that you can just start to get in. And and that's been like fun Jay Nathan: Yeah. Jeff Breunsbach: right now. Jay Nathan: Yeah. Very cool. Jeff Breunsbach: What else what else are on your laws? What else are all what what else are you thinking about? Jay Nathan: Okay. Well, there's one other thing. It's a little bit tactical, but I think it's pretty important. and maybe this is obvious to everybody else, and you know, sort of just became obvious to me over the past few weeks, is and it sort of goes hand in hand with software is free, everybody needs more software, not less, that humans interact with, but it's just bespoke. But software every every agent needs a Custom interface, right? Which is comes back to what we talked about with your CS platform. Because agents in the background, they can do a lot of things. They can process a lot of data autonomously. They can, we can leverage the intelligence that's available through these APIs to go, you know, process and summarize information, to do tasks on our behalf. But at the end of the day, we need to know exactly what these things are doing. And we need to be able to have humans in the loop at the right point in time. And that is basically. You're creating a harness around the agents to do that. So what you've created with your CS platform, in my opinion, is almost like a an agent harness, right? It's taking the inputs, it may maybe triggering agents, but it's also taking the outputs from agents and putting them in a format that humans can understand what they've done, learn from what they've done, apply what they've done, take the next action on what they've done if if they, you know, if there's an action for humans to take. Which there often Jeff Breunsbach: Yeah. Jay Nathan: is, especially still today. So every agent needs an interface and I don't think it's the chat window. Jeff Breunsbach: Yeah. Well yeah, because you st you still find yourself yeah, needing a little bit of like these areas of like where where stuff needs to happen, right? Or like where you need to be able to review certain things or to like Jay Nathan: Yeah. Jeff Breunsbach: you still find yourself needing these places to commune, basically, is the way I I've thought about it. Is like, okay, I need sure, I need a place to look at the output of some of these things so I can train it better, right? And like that to me, like I need I'd much rather just have an area I could go to where I know, okay, cool, I can go look at these I can go look at the results and I can easily pull those up and then go train it better on X, Y, and Z versus having to prompt the window to say, Hey, give me some give me the last ten and let let me help you. Like Jay Nathan: There there's a there's a lot of mental space between a generalized output and solution that I then have to interpret what this means in the context of a framework around a process. There was a lot of buzzwords Jeff Breunsbach: Yeah. Jay Nathan: there, right? So but what I mean is so Claude can generate artifacts all day long. First of all, they're not shared, right? They're well, there are some shared artifacts now, but it's not a shared application when I create an when I create an agent in Claude for the most part. It's it's my thing. So if I'm creating an output of files and saying all you need to do is just go re you know, go through this list of files and check them, make sure they're right. And if they're not, then you do this other thing in claw. There are too many mental leaps. So part of this comes back to part of this comes back to putting a shell around all of this information and intelligence that humans can understand and interact with in an in a simple way. Because Because we have more coming at us now, we have less time to interpret what we're seeing and make the translations. And by the way, I would guess 80% of the population isn't even pop they're not even it's not even possible for us to do that anyway. And I'll put myself in that 80%, right? Because we're just moving from thing to thing, we don't have time to think that much, which is part of the problem with all the outputs that we get from AI. They're they're too verbose, they're too much, they're not organized, they're not structured, which is where this whole idea of You gotta have custom software around the agents. You have to have UI around the agents. It's easy to understand, which gets back to design and all the product development principles that we believe in, right? You have we're gonna have more products being developed than we ever have. They're just gonna be bespoke to every every company, at least at some level. Jeff Breunsbach: I I have one minute and then I gotta go. But I think like I largely agree with all this. And the thing that I I've walked away with too is like more like you said, the team it's like you want your team actually focusing more on the right things to do, which means you actually want to limit the noise and notifications more than ever. But we have more software and we have way more ideas to create noise, right? But I think it goes back to this idea of like narrowing the scope and saying, Okay, sure, it's okay for us to create all this stuff. But at what point do I need to push this to a person in order to go act on something? And that to me is like a different barrier of entry now. We're like, okay, now, now this is the moment where I could go push this to basically the team to go action on. and to me, that is like a high that barrier to entry needs to be higher and higher because there's so much noise coming at people. so yeah. all right, this is a quick one. I gotta run. But I Jay Nathan: All right, ma'am. Jeff Breunsbach: think we should get no guts in here. Jay Nathan: All right, we'll talk soon. Peace. Jeff Breunsbach: All right, later.