Jeff Breunsbach: All right, welcome back to another episode of Chief Customer Officer dot IO. Jay, how's it going this week?

Jay Nathan: Good. another busy week. I'm headed up to Raleigh right now for a day and then headed home. Some partner meetings, couple interviews, and then back to Charleston. So how you doing? What's your week like?

Jeff Breunsbach: Nice. Doing good. Back at work. First, you know, thirty days of paternity is over, so I'm I'm back at work, so just getting in the swing of things. You know, yesterday was just I decided you you can tell me if I'm crazy or not. I you know, I I pulled up my laptop to I probably hundreds of emails and I mean we use Slack like consistently. So I would I would almost say like thousands of Slack messages. So maybe eight thousand, that's like you know.

Jay Nathan: wow. Yeah. Yeah.

Jeff Breunsbach: Like lot. But so I just went through and just nuked all the notifications. Just I just said, you know what? If if something was that important over the last thirty days, you know, somebody will come tell me about it and I just had Claude I asked Claude to go summarize like the most pertinent things and basically write me a I I told it a three to five page report on like the things that happened, the critical pieces that you know Would be interesting to a customer success leader, which product updates to be released, like that kind of stuff. and so I don't know if it's the right thing to do, but I I just figured, you know, I I don't think it's worth my time probably spending two to three to four hours like sifting back through all the messages and and stuff that you know, was that happened over the last thirty days. I'd rather just get a get a summary.

Jay Nathan: Most email is ephemeral. If you don't respond within a week, it's probably it's good to have as reference. I wouldn't delete it, but yeah. I've Yeah, yeah. I've I've never understood zero inbox people. Like I am I've tried years and years ago to be a zero inbox guy. N no chance. No chance for me. It's it's just not even I I could tell you horror stories about other things I don't pay attention to as well. I'm not gonna tell you all those

Jeff Breunsbach: Yeah. Yeah, yeah, yeah. Just I I marked them all as red, you know. Yeah.

Jay Nathan: 'Cause I don't want you to break out in hives, but like email is far from the top of the list of things that I pay attention to. So

Jeff Breunsbach: I mean the only thing that I do to email right now is I have a I have a t a label within Gmail that is just action. And so I will read an email and if there's an action it goes in the action folder and if I do the action then it goes out and that's it. And everything else just stays in the inbox. I mean it's all in the inbox, it's just like one label or not. And that just helps me like prioritize and now I have Claude doing that for me too. Like it basically reads the emails and sees that if I need to take action and then assigns the label. so yeah.

Jay Nathan: Yeah. Yeah. There you go. There you go.

Jeff Breunsbach: all right, I think there's a couple of things I wanted to talk to you about as I you know, I think there's one fun thing that I'm working on, and then I think there's like a you know, I'm getting back into work out of thirty thirty days of being out, and I wrote down three things that I thought you might find interesting of like me coming back, like what what I was thinking about. So the first thing that I think you find interesting is I have I've I think I've gotten my wife on board. I'm going to run our household with AI agents. Now that we have three kids, I was like, this is I was like This is becoming untenable. I have to now worry about like three dates of of going to doctors' visits. We have to worry about food f food prep and planning every week. We, you know, we have a nice home. I have to worry about changing out air filters and all this stuff. So I have spun up what I'm calling Mr. Baxter. Mr. Baxter is our chief chief operating officer of the Brunsbach household. and he is a he is a a harness basically for AI agents. And I'm building this on top with nanoclaw and I'm hooking it up to iMessage. I'm currently in the midst of buying a Mac Mini so that it can run on a Mac Mini so that it has iMessage capabilities because my number one constituent to get on board with this is my wife, and she will not do anything. She will not open a Slack thing, she won't open Telegram or whatever all these other ones are, like WhatsApp, like she's in iMessage all the time. So I'm like, okay, I gotta bring this to her. And so I've spun up Let me actually tell you what agents I've spun up and you tell me Let's see, this is bad, bad podcasting. I'll cut this out. But so now I've spun up, so it's not fully set up yet, but I've set up basically the like the outline. and the first one that we tackled was our home chef, who I wrote out instructions for in an MD file. And now what it has done is it's connected to our Instacart. it's connected to sources of like recipes, Instagram follow Instagram people we follow, websites that we have for recipes, like all kind of catered for my wife. I had my wife like talk out loud and I just recorded her talking about like I want you know, you know, I like protein forward, I like to have, you know, a vegetable and a fruit at every meal for our kids, I like to buy organic, whatever, all those rules. And it and it went off and you know, I had it run on Sunday and it went off and spun up a nice little website that said, Okay, here's your meal prep for the entire week, breakfast, lunch, and dinner. it came back with here's the recipes that you're going to need to cook.

Jay Nathan: Awesome.

Jeff Breunsbach: There's one day that we're going to prep a bunch of meals, which is on Mondays, and it's basically said, here's the things that you have to do, here's what time you have to start them in order for you to make a 5 p.m. dinner for your kids to be home, and all this stuff. And then it has synced into Instacart. So then I press a button and it literally goes and builds the Instacart. there's an API or there's a MCP that Instacart has. And so it'll literally go build our shopping cart, and I can literally go press, you know, purchase on those things. And so the idea is that over time, you know, this agent essentially can run almost like on its own via, you know, my wife texting it in iMessage and can kind of run in the background and basically continue to update its own rules. So I have we have a chef, I have a scheduler, so for like any of our kids' things, it basically is gonna go put this on the calendar for us. It's also gonna start to track things. So you know, when's the last time we went to the doctor? When's the last time we went to the dentist? anything that comes into our emails, it's just got view only access. So it's just viewing anything that has a date and understanding if it needs to be added to the calendar or not. so those are the first two that I've set up. but I've got kind of like a list of laundry list of those behind it, you know, kind of somebody who's taking care of the household. So when's the last time I got maintenance done on my AC? When's the last time that you know, I got maintenance done on our propane tank and whatever else. So, anyways, this has like been a fun little side project, but I like convinced my wife, I was like, Listen, like I want to go do this and like I think I could like basically get our time back from having to worry about these things and actually probably become more consistent than we are. Like I'll tell you right now, like I like you said earlier, like over the years I've like put it in my calendar, like change the ear filters, I've put it on put it in my notes app, I've put it on a text message. It never happens. And so I feel like if I actually had somebody that was looking out for me and actually just texted my iMessage and said, Hey, go change the ear filters right now, you know. I'll go do it. So anyways, I don't know if you think it's fun or not, but this is like a little thing I I started working on over the weekend.

Jay Nathan: Yeah, that's cool. So why would we be talking about this on a podcast about ostensibly enterprise AI, customer experience? Well, there's something well, I'm gonna tell you. So I wasn't asking you. So there if you haven't done this yet, it's a really interesting exercise because it it really opens your eyes to what agents are, which is an agent

Jeff Breunsbach: Yeah, yeah, go for it.

Jay Nathan: Is something that learns along with you. The cool thing about how nanoclaw, open claw, all the little claws work is they they're continuously learning from you. So you have your personal agent, which is what I would call this. And every time you you can you can coach it just by sending it a text, right? It's like, hey, you keep talking to me about this, but I don't want to talk about that anymore. I want you to do that. Or you you keep adding, you know, sugary things into the into the Shopify cart or whatever it is, the cart, Instacart. cart and I need you to stop doing that because we're gonna focus on vegetables and and it's just gonna do it right. You don't have to go reprogram something. I will tell you the you know we tried for years to do the calendaring thing in my house as well and it failed because it was all so manual. And I I like stitched together five or six Google calendars. I made a calendar for everybody so we could put everybody's events on their calendar and it would all show up on one and guess how many times that guy used exactly zero.

Jeff Breunsbach: Yes. Yeah.

Jay Nathan: but this this would actually this solves all that. So interestingly, I had this incredible call yesterday with a very large Let's just call it industrial services kind of company yesterday and and their AI team. And one of the pillars of their AI strategy is actually, and this is a company you would not expect to be as AI forward as this company appears to be. they really are leaning in hard. And and we learned a lot from talking to them, but they one of the pillars of their AI strategy is personal agents. for for some or all of their all of their people. And so personal agents within your your ecosystem at work will be a thing, just like they are in your home life. Right. And so anyway, I I just really encourage people, if you haven't tried this yet, go set up nanoclaw, go set up OpenClaw, whatever it is, and just understand what it's doing. It'll help you understand agents probably better than what you're building in house with Claude or maybe even with Glean. One more point. I had a call last week after the podcast with another company, about a thousand person company. And it was great. You know, he was showing off all the different quote unquote agents that they are building. The the problem that I see and the reason I sort of made this post last week about what's an agent versus task automation is that I see a lot of task automation happening. I don't see a lot of agents being built. Agents need to they need to remember and evolve and become autonomous on their own over time, right? And that's not just a technical thing like a a pedantic thing. That's like that's how you get the true value out of these. Because if you have to go babysit something, it's not it's it's not useful. You're not gonna go babysit it. You might for a few weeks and you're gonna realize, okay, this is too much work. I've got to get on to the next thing and I can't keep going back to this thing. And processes change every week. They evolve in slight ways, right? In nuanced ways. So sorry for the diatribe there, but I think what you're saying is cool and it and it does apply to like how we think about this for enterprise.

Jeff Breunsbach: Yeah. And I think like the the one example I'll give about how like it's learning alongside you is is the first time it went through and created like an Instacart, right? It just went and picked brands. Just, you know, like we kind of gave it some directions, but it just went and picked brands. And so we had to go back through and and start training it. Like, no, we like this kind of milk, right, Claire? Or like we like this we like the to do vegetables that are organic because of this reason, like blah blah blah. And so like you said, like now it's and and so if you go look at like the I'm I'm calling it like the soul file S O U L, like the soul file of this agent. Like essentially now you can start to see it's starting to layer out instructions for itself, like to remember and to like always call upon and and these things. And it's just doing that itself, right? Like I didn't tell it to get, hey, go update your files. It just now it's starting to do it on its own as we text it, right? Okay. And at the end of the session, it basically was like, I went and updated all these instructions so that next time I know this is what it's gonna be. And so now you like you said, like now you've got this thing with you can start to see, okay, now it's gonna continue to build on each other and and then, you know, like for instance, like this week it's like, hey, we're leaving on we're gonna leave, you know, on the weekend and so like we don't need a meal for this day. And it's like, okay, so this is different than a normal week. Like yeah, the normal week is is, you know, seven days or whatever. And so now it's you it's starting to learn, okay, you might have a travel day, so it's gonna ask you questions next, hey, are you here all seven days? And it just learned that, right? Like it just it kind of picked that up as like a nuance of something.

Jay Nathan: Yep. I've been using mine for just personal training and nutrition myself. Like we're not even using with a family. We're be we're sort of beyond that point now. and I literally like I'll t I'll just tell it or send it pictures of what I've eaten throughout the day. And it's like, okay, here's how much more protein you need, here's how many more calories you have left. You know, we know your goal is just to drop a few pounds and sort of get back on track health wise. So, you know, and then throughout the day I'm like, Okay, I'm

Jeff Breunsbach: Yeah.

Jay Nathan: I actually don't know what to eat right now. Just tell me what to eat. And it's like, okay, here's what you could probably do and it's I mean, it's incredible. So it's that that's been a game changer.

Jeff Breunsbach: Yeah. That's that is like the the I'll say like the number one reason I set this up is because of decision fatigue. Like I there is just I make every decision, yeah, I make a lot of decisions at work every day, and then I have to make many decisions now that I've got three kids at home. And so like you like you were just alluding to, like I literally just want somebody to tell me what to make for dinner. That is within our guidelines, within the things we like, and just like has a recipe pulled up. Like that is, you know, and then I want somebody to tell me when

Jay Nathan: Yes. Yeah.

Jeff Breunsbach: The next time my kids need to go to the doctor when they need to go to the dentist, right? Like I don't wanna go back and sift through, they went six months ago, like you know, like I want that stuff just pre planned. So anyways, that's that was just a fun thing that we're working on. and so right now I thought about doing it on a virtual machine, but then I kind of got nervous, like, man, then it just lives in the cloud and I don't know. I I'm like I'd rather I and it anyways is something else to tinker with. Like I'm just gonna go try and find a Mac mini. I think everybody's like trying to buy them, so they're like non existent right now. But

Jay Nathan: Yeah. Yeah.

Jeff Breunsbach: I just think it'd be fun, another fun thing to set up. Like, okay, I have to set this Mac mini up, it's gotta run on its own so that like we can then start to put these agents on it and it can run. and I'm sure there's other devices that work. It's just I've been reading a lot about that. So

Jay Nathan: Yeah. We just need something that's on and connected to the internet all the time. Like my my problem is I'm running mine on my laptop and so if I've closed my laptop then I'm like, shoot, I can't so it's not always on. I I'm I've gotta figure that. I actually was I was just thinking about going to buy an iMac just because it's a good excuse to have an iMac. Which I sort of want. So All right. Yeah. yeah.

Jeff Breunsbach: Yes. Yeah. Yeah, just having a yeah. Which is like the same thing, right? I it's just because Mac mini you just need to set up with a monitor or whatever or whatnot. all right, there's three quick things I wanted to tell you about that I as I'm coming back to work that I wrote down as like I guess like for the end through the end of the year, these are three things that I just am going to be maybe more meticulous about than anything. first one is that in this age of AI.

Jay Nathan: Yeah, exactly. Okay. Yep. Let's hear it.

Jeff Breunsbach: our human relationships are going to be more important than ever and we should we should treat those accordingly. And so what I mean by that is that like if we have a meeting with the customer, if I'm sending an email, if I'm doing anything, like I like we need to have human personalization. Like we need to have a human on that. Like I don't want to just automate everything or AI everything to death. Like I actually think in this age of AI that like our relationships with customers, we could actually like I guess like almost like counterpoint to AI. Like we need to be investing in those relationships even more heavily. And that's where I want my CSMs spending time is basically like automating anything that doesn't work or like doesn't work for you and leaning into the relationship side. And so to me, those are things like how do we surprise and delight customers? Like I think that's kind of gone out the window, but like how can I how can I learn something about a customer, send them something in the mail, you know, send them an article that they might have missed, you know, anything like that. But I just think like surprising and delighting customers has gone out the window. I think the other thing I wrote down was just like

Jay Nathan: Totally.

Jeff Breunsbach: Sounds silly. I think I've been beating this drum for like ten and you're probably annoyed at this, even if you remember. But like meeting preparation I still think is like for the birds. I think no one does meeting preparation anymore. It feels like every meeting that you attend is always like, Okay, so what's the agenda? You spend the first five or ten minutes, okay. So what are we trying to figure out? What are we doing? Like, man, could you imagine how much of a difference we'd make if we walked into that meeting and we already had an agenda, it was crisp, we walked in and we actually like took meeting preparation like seriously. Like again, I think that's just another example of like how you could make a difference. So it's like Number one that I wrote down is this whole idea of leaning into human relationships. number two, we have to be maniacal about eliminating internal internal processes. Like we have to be, I I was reading an Elon Musk thing about how he he always tries to remove stuff. Like if I look at a process, he's always trying to figure out like, how can I remove anything? And I think we just have to be we we have to have that mindset of like map the process, take the time, but then be maniacal about like Does this pro does this step need to exist? And if so, you know, is it human? Is it agent? Is it AI? What is it? and if it doesn't need to exist, can we eliminate it? But I think we have to be really maniacal about that. it just to me, like, I I think sometimes we just say, this is how it's always been done, we we just need to do it. I think it's just pervasive. So I think like being maniacal about removing steps in the process is another. and then my third.

Jay Nathan: Have you have you ever seen this right here? Have you hold on a second, I wanna show you something. Just to have you ever seen this picture of the so the the different generations of the Raptor engines at SpaceX. Have you seen this picture before? Hold on, I'm gonna pull it up and then I'll describe it for everybody. How do I do it?

Jeff Breunsbach: Maybe. wait, I think so. Isn't it like something that looks almost like a a skyscraper to begin with and then like every generation yeah, yeah, okay.

Jay Nathan: So y yeah, it's it's like it it looks like so generation one is literally a mess of tubes and pipes and wires and it looks like it was built in a barn, like in the middle of nowhere. And then the next one is just a little bit more streamlined, and then the third one, it's like where did all that stuff go? It's just like two pipes. It looks like a very streamlined, incredibly, you know. efficient form factor of an engine. And I think it just speaks to like, you know, Elon is sort of famous not just for for, you know, cutting all the process out, but also making the product iteratively better. And if you could do it with a big manufactured durable good like that, surely you can do it with your processes within your company. Right. I mean, so anyway, just to interject and and and throw that in, look that up if if if if If you're listening to this, you'll be sort of astounded by the by by the way that that thing has evolved. But it's to your point, Jeff.

Jeff Breunsbach: Yeah, the third one's almost you'd almost call it beautiful, you know? You'd almost say like that's a pretty it's a like it's honestly like a a beautiful looking product now, because of like its design. They actually thought about that. It kinda reminds you of the I think there's the I think there's a Steve Jobs quote about him spending time on the package design or something, right? In the or even on the inside of the device. I think that's what it was. And and yeah.

Jay Nathan: Yeah, yeah, that's right. Yeah. And maybe that's the way it should happen. Well, on the inside, yeah. Yeah, like even carpenters, right? They put they put nice wood on the back of a of a cabinet, even though you're never gonna see it, right? Because it's craftsmanship. But I think the difference between what Elon does and what Steve Jobs did is that Steve Jobs would try to make the first version of the product look like the last one. In in Elon's world, right? Elon was just getting it done as fast as possible. That's part of the reason it looked the way it did. Like he had a problem to solve. But then he refines over time to make it beautiful. So I think there's a little bit of a contrast in style, you know, execution style there. But anyway, neither here nor there.

Jeff Breunsbach: Mm. Makes sense. and my third my third one is is to bring a little bit more joy and levity into into our world. Like I think I in particular maybe don't celebrate some of the moments enough, you know. I'm just kind of focused on the next thing. I've you know, it's like a again, I've got kind of a busy home life and so I just kinda like log in, all right, let's let's go, let's move, let's do it. And I think I need to try and figure out more ways to enjoy the ride up, right? Like I'm I'm currently at a business that is growing very rapidly. It's a very exciting time. Like there's a lot we're trying to hire, you know, roughly thirty people this quarter, like across the business. Like we're and we're, you know, it's gonna basically double the size of our business. Like we're we're moving fast and doing a lot of cool things and in like I don't know. You're it it's almost like you know, I I wanted a seat on the rocket ship and then like once I'm on the rocket ship I ha I I haven't like looked outside or something. I don't know the right right verbiage, but I I think that's just another one that we've got to try and figure out how to

Jay Nathan: Well. Yeah.

Jeff Breunsbach: But to bring more into like the day to day world.

Jay Nathan: Totally. Yeah. I mean, you're getting some great experience there. with sort of a hyper scaling kind of company. I mean, it's it's very cool. So I like those, Jeff. I think that's good. but you know, I think the the trick of number one, which is the, you know, focus more on the human stuff. I think you've got to get really good at automating for the ninety seven percent of of customers and

Jeff Breunsbach: Yeah. All right. Those are my three.

Jay Nathan: interactions that that need to be automated that you can't afford to you have to really pick who you're gonna go invest in with with your human time, right? Where can you actually make the biggest difference? Right. So it actually forces you to do both the the scale stuff and the and the high touch stuff really well. So that's why I like that one.

Jeff Breunsbach: Yeah, and it it honestly it's I think it's forced us a little bit, even in a short period of time to to again like visualize more things. Again, I think we we're kind of moving so fast that you you sometimes you just kind of think, okay, we already know what the process looks like, we already kind of know what this customer journey looks like, and you kind of just skip it. you know, just kind of moves on. And and so I think like there's a

Jay Nathan: Yeah.

Jeff Breunsbach: there's a piece of that that like we just need to slow down in s in some cases to speed up. All right, enough about that. what do you want to jump into next? You wanna you wanna jump into this for deployed engineer article?

Jay Nathan: Sure, yeah. So the Ford deployed the ten billion dollar Ford deployed engineer boom. I think we talked about it last week, didn't we? It was seven point five then. AI companies are embedding engineers inside your customers. Nine point seven five billion committed in twelve months across Amazon, OpenAI, Anthropic, Microsoft, Salesforce, who evidently is adding a thousand reels. Interesting anecdote about Salesforce in a second. I don't know if I shared it last week. Google, one quarter of Accenture's annual labor costs going to FDEs. So really interesting. It seems like there's a couple of different models that these FDEs are being deployed in. One is like they're making actual equity investments. So Amazon and Microsoft are doing joint ventures. Anthroopic and OpenAI are are doing private equity funded. Entities and then Google is mobilizing the partner ecosystem through systems integrators. so it's just the everybody I talk to. So I had a call with a friend of ours, Jeff, last week, who is with a large VC firm. They have hundreds of companies in their portfolio, and they've been doing deep, deep research on FDE. And I mean, all of these companies are. investing in FDEs now. So what is an FDE? you know, I think we've talked about it before here and we continue to talk about it, but FDEs are for deployed engineers. They're actually so there's this huge capability overhang, meaning the models, the the tools that we have available to us, they are all way more capable than what we're implementing today.

Jeff Breunsbach: Mm.

Jay Nathan: Or th than what we're able, what most companies are able to implement today. There's just so much technology now, so many tools available. models, harnesses, application layers, all kinds of stuff. So and right now, very so here's a really interesting observation. I was on a sales call last week with software company, and it was very I just had a really Big awakening. It like the sales rep on this call was asking these loaded questions of the client, like, hey, do you have this problem? Do you have that problem? Do you have this problem? And why do you think that was, Jeff? Because he was trying to find, he was trying to find out whether they had the problem that he could sell a product for. And I feel like it was such I don't know. For me, it was such an odd thing to watch because

Jeff Breunsbach: Yeah.

Jay Nathan: I think what everybody is asking for right now is not they don't want your product. They want a solutions. They they want end-to-end workflows built to completion. And that is why services is having this boom moment. And the main delivery model behind that today is for deployed engineering because this stuff is if done well and done right and actually done at scale is actually somewhat complex because of all the technology that's in the mix. You've got application vendors that have their own AI components. You've got platforms like OpenAI and Anthropic that have their own tooling. You've got you know the hyperscalers, Azure, AWS, GCP that have, you know, tools that you can use to go build agents. And then you have agent platforms like Glean and Sierra, these other sort of open, they might be have a little bit of a major in terms of the problems that they solve, but they're still sort of open platforms. so putting all that together in in a into a solution that actually makes sense is a little bit of a daunting task. And not everybody has an AI whiz kid internally who's helping who A has the know-how of what all that tech does and how to put it together, and B has the organizational in intelligence or organizational clout to go organize the company around deploying those kind of solutions. So anyway, FDEs are one logical extension of that. We're actually going all in on that. Our our our company is, you know, we're we're hiring FDEs right now. We have FDEs on staff. We're, you know, that that's the the work, agent deployment work is what we're doing now, but there's so much underneath it. So anyway, sorry to monologue again, but just

Jeff Breunsbach: No, it's well I I think the thing that I'm curious about if you you know is is it I think if you if you kind of think back to I'll call it like the consulting days, you know, like consultants would go in and and they would try and like deploy they they'd almost parachute into your business, they'd learn about your business for a couple of weeks and they'd like basically come back with some recommendation. Okay, I interviewed a bunch of people, here's all the stuff, right? Like and they and then they'd kind of parachute out and leave you with a bunch of deliverables, right? And I think like this is kind of akin to me for a to what a forward deploy engineer could do. But to me it it actually I guess layers in not only the discovery and the strategy, but actually layers in the execution, right? Because if I can leave you, if I can not only leave you, but maybe even stay with you and continue to evolve these agents and all these other things, right? Like now you've got something that's actually being executed in your business. And so it's like this person comes in and I think like the the I've heard

Jay Nathan: Right.

Jeff Breunsbach: a couple of different interviews, so I'm I'm mixing them up right now. but I wanna say it was either Stripe or somebody some one of the one of the leaders ha has come out and basically said like their organization is moving to a structure where it's essentially they're pairing up somebody who has the business context or the domain expertise with an engineer and like they're trying to make these mini pods. And so it's almost like engineers are actually making their way into other parts of the organization now.

Jay Nathan: Yeah.

Jeff Breunsbach: Like you said, to to basically bring okay, you know, this finance person can bring a layer of okay, how do I close the books every single month? And what are the steps that I do and how do I do that and who are all the vendors? And all they can bring all that stuff. And then the engineer can look at that with a an objective sense and say, cool, I can go build a harness that, you know, looks at those steps and we can optimize our agents and we can, you know, I don't know, Geminize great with numbers and so, you know, let's not use this and you know. blah blah, whatever, right? But that I guess like pairing pairing those things together is where you get, like you said, like where the I guess it to me it it almost feels like it's a that execution layer that like has been missing for and even for a lot of not only consultants, but like for a lot of like SaaS businesses, right? Like if if I I don't know, you bought my product and nine times out of ten, right, I I was just basically hoping that you'd I I could help you along the way, but like there's like that last lens of like, okay

Jay Nathan: Yep. Yep.

Jeff Breunsbach: I actually don't know if you went and made the changes inside of your business, inside your organization to actually like see that through. And now this is like that last almost like that last leg of the trip.

Jay Nathan: Right. It's The last mile. That's exactly right. The last mile is the hardest mile as well. Right. yeah, I think we are it's probably already evident to people, but it we're in a battle for talent again. And it's that kind of it's this kind of talent, FDEs. The the thing that you just said is really interesting because you know, I think the canonical definition of a of a forward-deployed engineer that that people want is somebody that can do a little project management.

Jeff Breunsbach: Yeah.

Jay Nathan: product management, that's an engineer, that that can also be have some depth in a business domain. The reality is, those are gonna be unicorns. They they are unicorns today. and so I think the pairing of the business operator with the FDE is a really important kind of thing. In in fact, the company that I was talking to yesterday,

Jeff Breunsbach: Yeah.

Jay Nathan: Again, $10 billion publicly traded. They're sort of running ahead of, you know, where I see most, especially companies in their kind of industry in terms of AI. What they're doing, you know, they've they've built out the the AI architecture within their organization. And now what they're doing is taking subject matter experts out of the business and training them on AI. Because these tools aren't and we're not talking about technical necessarily. Once that architecture is set up. We're not talking about technical anymore. We're talking about taking business knowledge and and putting it into an agentic platform, right? Building agents by sharing what you know, connecting to existing data sets. There's still technical work to be done. You still have to surface the right data. You might have to, you know, build rag systems for retrieval. You might have to do model optimization. Some of that's gonna be technical, so you're gonna want that, but they're doing it the opposite way. They're actually Creating forward deployed engineers by teaching subject matter experts in the field the the pieces that they need to know about AI.

Jeff Breunsbach: And so then they and I guess their hope is that like that person who's probably already got some level of domain expertise, if they train them up on the the AI components, then becomes even more powerful because it's like, okay, now now you know how to operate within the confines of what we're building in your own domain. And then basically you become the I don't know, the keeper of your own future, right? Like you've got a little bit more of your because again, I I think like one of the I

Jay Nathan: Right.

Jeff Breunsbach: Maybe one of the challenges that I'm coming against right now in our business is, you know, we're we're small and and trying to move fast, but still, like you said, I think the infrastructure, and and you know, we're in a healthcare industry, so we also have to think about some data compliance. We have to think about like systems that are connected to certain things, like that that actually requires my head of engineering and like some of our lead engineers to help me architect that. So I'm still I'm until that infrastructure is set up, right? I'm still on this little bit of a hey, I'm on the I'm on the roadmap. I'm waiting for my I'm waiting for my thing to be built that I need, right? and so I think like you said, like if you can if if they can build the architecture and train them on how to use it, then you know, you kind of put the fate back in their own hands. It doesn't become like a another an another pipeline that maybe doesn't get to another task on a pipeline that doesn't get built.

Jay Nathan: Yeah, exactly. That's right. Another roadmap that doesn't get fulfilled. Right. I mean, but that that is the great equalizer here, the great, great leveler of playing fields is that now, if so many more people can can go implement the things that they need, even if it's an application, right? We could teach people how to build the applications that they need. Now, there is an element of depending on how many people are going to actually use that application.

Jeff Breunsbach: Yeah.

Jay Nathan: You might want a little bit more fidelity to that. You might want to do enablement and training around it. So I think all the traditional roles that we've had in in terms of services and software deployment and technology deployment are going to exist in some shape or form, maybe in different measures. right now, especially. But yeah, all those things are are still required to go deploy these solutions. So

Jeff Breunsbach: Do you I'm I'm curious if I know you've had a lot of conversations in the in our last couple of minutes here. I'm curious. you know, it's still or or maybe do you th I guess do you see this evolution of us talking about we kind of talked about AI or I mean like the industry as total, right? Like kind of AI came about, everyone used it personally, then it changed into okay, how do I optimize this for business? And now it's kind of turned into like agents. And so do you I guess like do you see like agents right now, I guess as like those those people who are building agents are the early adopters. like right now, is that what kind of where we are? Is that like there's a a small number of companies that are really doing this, I guess like doing what is being talked about, but like those people are going to get a massive early mover advantage.

Jay Nathan: yeah, I I think the the jury still remains out on that, on like will it actually be a massive advantage for those companies? it's still very hard to measure. one of the conversations we had yesterday was well, it's hard to measure and there's this this big looming question over over Like where the intelligence is going. Are you giving away your trade secrets as you're building this stuff? so it's called, you know, I guess the terminology is called intelligence sovereignty right now. So if you look at what the big model builders are doing, you know, they're they're starting to launch vertically focused businesses and products. Of course, code being number one, right? Which all the code in the world was out there open source, a lot of it, open source helped, you know, these models build from s from the start.

Jeff Breunsbach: Yeah.

Jay Nathan: But as you, you know, as they learn from what you're you're putting into them, you know, they can they can provide that intelligence to your competitors. So or they can build it themselves, right? Whether they're doing it intentionally or not, it it it's it's bound to happen. although, you know, that's where sort of the enterprise level agreements with some of the model providers come in. You know, don't train on our data.

Jeff Breunsbach: Yeah.

Jay Nathan: But there's there's growing w one of the convictions that somebody we talked to yesterday had is that the open source models are gonna be, you know, three months behind the big, you know, frontier models probably forever at this point. And they'll probably catch up at some point as well. And as long as as you have access to open source models, which everybody will, they're not gonna get kneecapped like Fable Five has been. And if you don't know what I'm talking about, go read about

Jeff Breunsbach: Yeah.

Jay Nathan: You know, Fable Five and the fact that it, you know, it it decides whether it thinks what you're doing is appropriate or not. And it will basically nerf itself if if it doesn't like what you're doing with it. Like, how's that gonna work for enterprise? Anthropic has got a real problem on their hands. and then, you know, the the the cost is the other piece. We always come back to the cost, but you don't need fable five, which has intelligence Over, you know, every piece of context and text that it could find in the on the face of the earth. Like you don't need to run multi-trillion parameter models to get very specific jobs done within an enterprise. And so post-training models is is something that you know Microsoft's AI Foundry is trying to make very accessible and easy. So you can take your data, you can couple it with an open source model, you could sort of blend those two things together and create a model specifically for your business or the tasks that you're trying or the the kind of jobs that you're trying to get done with AI. Much lower cost. You don't have to worry about your your IP leaking out. And you know, you're you're basically owning your own stack at that point for A AI inference.

Jeff Breunsbach: Data. Yeah. What's interesting about that too, right, when you think about Microsoft strategies that they actually own open AI, right? So they've they've kind of got both sides of the market, in in a sense, right? Like if they're gonna do this open source stuff too, which is smart. Or at least in from the the outside they're smart, right? Because you're gonna have enterprises wanna, you know, have control and they want to make sure that like you said, they can keep their advantages where they are. So well yeah, that's cool. Yeah, I

Jay Nathan: Exactly.

Jeff Breunsbach: I think the whole forward deployed engineer stuff is is super interesting. I I to me, like the the the big thing that I've I don't know, been reading about and and stalking videos about for like the last six months is this whole idea of like engineering, you know, how have engineers worked, have product managers worked. Like I just think like those I don't know, the the those types of people have always thought about solutions and how do you approach problems in the right way. And I think like that thinking now is is gonna be really if like you said, if you're on the business side, if you get that if you get that style of thinking, if you're starting to write down problems in the right way, you're starting to articulate like the things that you think are important in a way that you could then go build AI and and solutions and agents and things around. Like I just think there's gonna be such a I don't know, there's I think that's gonna be a massive advantage for somebody who's in a business right now, is like go learn some of these skills and and kind of the I don't know, the mindsets and the methodologies that engineers and product managers have had.

Jay Nathan: You just you you just nailed it. I I've told this story before, but I I always used to have a I had an engineer that I worked with for many years and when I was in product management and he was always so good about coming back to mu much to my chagrin at times, but coming back to the question of like what problem are we actually trying to solve? Right? Let's get real clear on that. And I think you just made an excellent point, which is In the age of AI, like getting people to think about not what they could build, but like what problem it is they're trying to solve, that's the skill. That's the skill of the future. to get everyone to think in terms of business, like the the data structures, the entities, like the the kind of objects that they deal almost to abstract themselves away from them their job and think about what it is that they're working with right. It's not everybody's going to be able to do that, but that is the skill of the future in terms of being able to sort of build companies, operate, you know, them autonomously to some degree at scale. So I know there's a couple more stories we wanted to get to. I don't don't know that we're going to get to them, but that there's a Starbucks story that's very interesting. We should talk about it. Maybe we save it for next week. in our

Jeff Breunsbach: Yeah.

Jay Nathan: Our podcast agent will probably pick up on that and tee it back up for us. So, Mav, do that.

Jeff Breunsbach: Yeah, let's do it. awesome. All right. Well it's good to talk to you. I like hearing about some of the stuff that we're building and also I think it's been super interesting 'cause I feel like you're having conversations on the front lines with companies and sales calls and whatnot. So I think there's it's interesting hearing some of the market insights that you're learning.

Jay Nathan: And so some of them Some of them were ahead of the curve significantly on. Some of them blew our mind. so the call we had yesterday was was pretty pretty impressive. We were really impressed with what they were doing. so anyway, we'll I'll we'll keep talking more about that. Good stuff. All right, man. Yep. Have a great week. See ya.

Jeff Breunsbach: Cool. All right. We'll see you next time. Thanks, Jay. We'll see you, man.