E54: Become AI Native with Jim Benton from Adapt

I guess the thing that I'm most curious about, because Jeremiah, like, mentioned organically in the, in the last conversation that at this point it's like the bill of his that's going up the fastest because of how much it's getting used inside of the company. And so the, the, like, the meta question that I have is, do you, do you find the way that, and, and of course you can like quickly describe the product, but I think the bigger learning that I'm most curious about for other SaaS operators is, or software operators, Um, is do you find that it is significantly more important to have that expansion, those expansion moments versus the land moment? And then how do you organize your company internally to mirror which side of it do you believe to be more important? Yeah, I think, uh, hey Jeremiah, good to see you on that. Yeah. And actually, Jim, can you jump in too and just tell, tell everybody a little bit about Adapt and, and what it is that you do? Cause that'll, that'll help with this answer too. Sure, sure. So Adapt is essentially your company brain, but it's built and powered by the Adapt AI computer. And that is a very fundamentally different part of how we've approached things by being an AI computer versus just an LLM call, uh, and using MCPs. And I think that the, you know, the key is once you connect all of these tools, so we'll see companies connect, you know, they'll start with 5 to 10 tools. Next thing you know, they've got, you know, 20+ integrations, GitHub and Linear and HubSpot and Google Analytics, you name it. Once you put those together at a team basis, it's this pretty insane thing where this this computer has access to all of your knowledge. It can take action on the tools. It's all at the API basis. And so what you find is that people can know anything, do anything, and now it's just about how do you leverage that computer to do work that you just never thought was even possible before? And so I think to answer your question about what's more important, landing or expanding, I think in the AI space, I mean, it is busy right there out there today. I mean, we see it, right? It's a fast space we're in, especially coming outta the SaaS days. So landing's important. You gotta break through the noise. You gotta find that thing that's different, that resonates. And I would say the big thing is just being frictionless. You know, the old SaaS approaches of heavy salesmanship and, you know, click to sign up and talk to sales and go through the full process. The best AI companies we see out there are the ones that are moving really quick, reducing the barriers, getting people right in on a self-service motion, and then having like wickedly strong sales assist motions to engage at the high levels, senior level CEOs. Um, but again, staying frictionless with them until we get those first types of tools connected and then we can hit on the expansion. Okay. Okay. So I'm gonna try to say back to you like roughly 3 steps that you think are, it sounds like you have learned are really important. One is you want people to find that first magical moment without necessarily talking to someone on your team. When you, 'cause you, you, you said a few things. You said self-serve, you said they start with maybe 5 connections. So there was a few things you said that kind of triangulate to when they see that, when they hear about a use case and when they set it up on their own, they have a starting point where they get that aha moment on their own pretty quickly. And then, then, so, so now I have a question because you said Sales Assist. Yeah. Do you ask your sales team upon that aha moment to email the CEO or the user who got the aha moment? Like, like, yeah, help, help me understand what this, where the step 2 happens from that first moment. Makes sense. We went GA last month so that we moved from sort of this early access mindset where we were with design partners. You would sign up, our sales team would reach out, we would do a Zoom call, and then we would connect you literally live on the call. So in the 30 minutes, we would have you up and running and connected, which is powerful. I think to go deeper with GA, yeah, CEO can get in. I saw one happen this weekend. CEO got in, connected several tools over the weekend, started to run different queries against them, and sort of got to their own aha moment and put a credit card in before sales could even reach out. And then at that point, sales is able to reach out and sort of paint the bigger picture, really talk about an org-wide piece. But that, yeah, that is a massive change for us to get somebody in literally connecting linear, you know, I think it was Notion, uh, and different systems without talking to sales. That was a moment when I came in Monday that was like, we are, we are, we are doing the right foundational pieces to be one of the great AI companies out there. So Jim, I'm gonna do you a favor. Yeah. I could use Stavers. Because, because your, because your sales team won't love to hear it directly from you, but CEOs play with stuff on the weekends. Like, that's what we do. That's right. That's our playtime. And so you're gonna like, and, and, and so I don't know that you will see this consistently, but maybe you will see over weekends, just like you just saw CEOs playing with your tool and then having that magical moment. And now the sales team is gonna reach out on Monday. Yeah. I just wonder what's different. It's like, it's like, it's like, it's like in the restaurant industry, you would never have restaurants closed on the weekends. Yeah. It's interesting. At the end of the day, I don't know if the CEOs are really looking to have a human reach out to them at, you know, 10:00 PM on Saturday night in the old classic SaaS way, right? Where it would either be a sequence or kind of a lightweight, hey, saw you signed up. That, that, that era of is over. I think the, the human era is not, but just sort of the generic, just a fast, quick touch of like, hey, we're here. So I do think that getting them in and getting Getting them to the aha moment in a self-serve way is such an unlock for an AI company. And the, the great companies we all know did that. I mean, think about the first time you used Cursor. Nobody reached out, you literally downloaded it, you did a few things, and within an hour you're like, holy crap, I have to tell every friend I know how cool this is. I think that's the experience. Now for us to get the aha, just using Adapt to do a search similar to how you might be using, you know, Claude or ChatGPT, although we actually do really remarkable things on that, that's an experience most people are used to, which is I didn't connect it to anything. I'm gonna do a search. It'll look similar unless you really push us hard and leverage the computer aspect. It's when you connect a tool that we get a very big aha moment. And so connecting Slack is maybe a first use case and then doing deep analysis of your conversations and then you connect and then you quickly realize with a couple of clicks I can connect a couple more tools, maybe pop an API key in. And that is usually the quick aha. They ask something and they, you know, it'll cook and chew and 10 minutes later they're getting something that they just have never seen before. And at that moment We've got a customer. Yeah, it's interesting. I mean, just to touch on this as somebody who is a customer, I think that that was the big thing for me is the ease of integrations, which actually, Jim, at some point I think, uh, I want to come back to this, but integration piece is really compelling. And one of the things you all have done, and I, I'm curious, this is something I wish I could replicate. I don't know how to do it, but eventually, uh, the, like the way integrations work is you literally take an API key, uh, and you give it to Adapt. Adapt, there's some like, some things like a Slack Connect, but a lot of, like, a lot of the tools we use are smaller companies with their own APIs. And so we basically really just grab the API key, give it to Adapt, and Adapt will go figure out how to connect it and get the information, um, which is incredible. So my team of non-technical people, like our technical people use, or by, and by technical, I mean our engineers, they're all using Cloud Code. They're not actually, they basically don't use Adapt. Adapt is used by our, more of our customer-facing and internal teams. But in that context, those people with no training can actually go in, give, give Adapt an API key, and Adapt just goes and figures out how to pull the data, which to me, that was one of the biggest things is within minutes, basically you have all these different things connected and you can start asking questions in Slack. And then for me, what I did with my team was when they would ask me questions, once I got this stuff connected, I would just say, hey, Adapt, and then give it the question, effectively a different version of it, and show the team like, look, you can just get this. And so now the behavior in our company like very quickly has shifted. It's not 100% there, but has very quickly shifted towards that being the default where people actually just ask the question. Um, and because it's so easy to connect to the integrations, it's been really easy to also drive the adoption. And so I think that's, to me, that's actually, I agree with that as like the, the thing that's most compelling. Um, and I'm curious, like how, how did you come to, I guess that, that way of doing things? Because I think typically what would have happened in the past is you would go in and there'd be a whole process. You would need a custom integration, you would need all this kind of stuff. Yeah. But you've been able to say, just give us the API key and we'll figure it out. Yeah, I think it was an aha moment. It was a little bit of an aha moment in the early days of, of building Adapt. One of my, you know, co-founders, incredibly strong technically, had a, had a lot of experience before Adapt, both on the database side as a founder building just a wickedly fast database, but also in the sandboxed containers and spinning up, you know, virtual machines for, you know, past work. And so when you, when he thought about how do we enable AI to have these special powers and to do incredible things with APIs by giving it its own computer, it means that the, the AI can figure out how to make sense of that API, or you could give it its own instructions and adds it to the knowledge base that everyone in the company now has that. But essentially AI knows how to make sense of these API keys of all these different tools out there. It can write custom scripts to make it work. It can write the custom code. And so instead of us pre-configuring thousands of integrations as the old SaaS model would've been, right? You, you go through and the engineers build it and it's very, very structured and rigid, we're able to give AI the key and say, what can you do with this thing? What's all the magic you could do? What custom code would you want to write so you could do exactly what the user wants? And every time the user asks a question, whether through Slack or a web app, it'll write that custom code to make that sense. And most importantly, it can spin up lots of these different, you know, swarm of different connection requests at the same time. So unlike an MCP, like a cloud where you're going to go through one request at a time, it'll schedule a tool call, then a tool call, then a tool call. We'll swarm out and say, hit Linear, HubSpot, grab something from Notion, and then pull the data back together in a very precise surgical way, then run Opus, you know, 4.7 against it. And that's the magic why people say, we've just never seen an answer like Adapt. Like, how is it that you're generating such a different answer than what we're seeing even when we use, uh, Opus 4.7, you know, directly in cloud? And that's the reason. You give it a computer, it's writing surgical pieces, it's not having context issues, and it's not Scheduling one after the other, it's swarming. I gotta ask, I imagine this is something that a lot of people building in specifically in e-commerce SaaS are hearing and thinking, oh, that would be nice. Cause they just set up the Meta MCP or CLI and they're like, it doesn't work, right? Or they set up Marketing API and they get their account banned. I'm like, why? Right? That's the number one question I just keep seeing on X, my whole feed. Why is my account banned? Right? I set up the integration wrong and now I, you know, my ad account is banned or something. So integration stresses. Probably something that a lot of people can relate with. But why, why did you call it Adapt? Like right now, it, you know, I imagine there's a few people like me that are like monkey brain listening to this and going, wait, I still don't get it. What, why Adapt? You know, good integrations, good context from all my different apps for the LLMs to use. But like, what is Adapt? Like, what does it actually do? And why would I sign up and get that aha moment from Adapt? You know, my side, after we, um, after Chorus.ai was acquired back in '21, I was CEO of Chorus. We were leading AI pre-LLM. Motion. I spent the last few years down at Stanford StartX. I'm a mentor, lead mentor for the Stanford StartX accelerator, working with just a lot of great teams. And I was building myself, so I'm in it. I'm with these young and, you know, more seasoned entrepreneurs. And I do remember, you know, about a year and a half ago, it's just very early, one of the teams across from me mentioning, are you using Cursor? And I wasn't. I think our team was using some Webflow and some other, you know, kind of previous generation tools. And a friend I was working with at the time said, let's, let's get you into Cursor, especially as a non-engineer. I've always been leading a lot of great product teams and close and I love the front end, but I got into Cursor myself with the guidance of like a CTO that it's, we're doing production work. It wasn't just, you know, vibe coding something. It was, it was tied to an old legacy app. And within 24 hours I had that moment of, wait a second, there's two AIs out there. I had no idea that the developers had one AI that's the greatest product market fit I've ever seen. And everybody else I know and see at social events is using a completely different AI, like a ChatGPT experience, which was special. And remarkable, but nothing like the AI that developers had that actually did work for them. And so I had this moment of, oh my gosh, the world is about to fundamentally change in like a 12 to 18 month period and no one knows it on the business side. The developers know it. And that's what, when I, when I connected with Sean and John Andrew and the other two other co-founders, I saw the early roots of what we were doing. I was like, we're about to bring the other AI, an AI that can not only understand but take actions for people, like do the work to every other user in a, in a completely natural language way. So not, you know, detailed UX work streams for agents. No, just talk to it. You wanna build an app that sits on top of HubSpot? Just talk to it. And you wanna do incredibly complex things like scan every sales call for the last, you know, year and build a product roadmap based on customer feedback. You can't do that with an LM. You can do that when you have a computer. And that's the part that led to, wow, we're about to get— make everybody AI native. And Maybe to finish that point, it's so important. Like, I, I, we've had fun in our career, especially in the SaaS days, a lot of fun building these things up. We're here in San Francisco, we're in the Presidio. I mean, we have front row seats to this. If you don't go AI native soon, if your leadership doesn't help your team, you know, the teams go AI native. Personally, I'm really sad thinking about the outcome of what that is. I mean, just think about your friends, your, your parent, whoever it is. If they don't go AI native, which means they just don't embrace it, they don't learn to get the superpowers. They will get pushed outta the workforce. They're not gonna get hired by the new AI-native manager. They're gonna retire early. And so just think about sitting around the dinner table with people that have been marginalized outta the workforce. They're kind of bitter now on AI. They're, they're young, they're too young to be retired, and they've just killed their earning potential. And so if we could solve that with just a click, you know, Jeremiah, you mentioned just click of an integration, everyone in Slack had your tool. So it wasn't that they had to go install it. You put it in once, they said add Adapt. We're now instantly making everyone AI native and literally the next day they're going home saying, I've never had so much fun at work. They're having a cursor, Claude Code-like moment. And I mean, what a gift, right? That's why I keep telling people work has never been more fun, but we need to get these superpowers to you. Uh, otherwise, you know, we're not doing our jobs as leaders. I really like the name now. Okay. There's a bunch of things that I would love to dig into, by the way. Let's do it. The most important one being I love John Andrew and did not realize, uh, he was your co-founder. Oh yeah, brilliant. Brilliant. Yeah, yeah, yeah, yeah. I, I don't know how I did not make the connection prior to literally right now. And he's the source. I mean, the initial, just to play on that, John Andrews, a solo founder of Wander, Thiel Fellow, very unique, special person. If, if, as you know, meeting him, uh, just brilliant. And as a solo founder, I mean, it's, you know, running Wander, he wanted a better way to use AI to run the business. And that's the early origins of, wait a second, what if we built this thing called Adapt? He pulled in,, you know, an early engineer he'd worked with, you know, very brilliant, uh, engineer. Um, I got connected in and was like, this is exactly the future I see, especially this blank canvas kind of UX approach. And that was the early origins. And then we took that out and did our seed round and so forth. That's amazing. Yeah. Um, one of the things that you said, first of all, I, I, you can push back on this, but you said something like early retirement and I want to push on that. Because I am not sure that age is correlated to whether or not people are becoming more AI native. And the reason is I am seeing this very weird phenomenon where candidly, like roughly people my age seem to be the most resistant. So I'm like, I'm 40. I also see a bunch of people who are resistant who are like 23, weirdly enough. And conversely, I've like shared this story before, but my, My mother is like embracing it big time. Amazing. Amazing. And I, and so I just am not like, and maybe, look, I have no idea what bias my sample has, but for the first, usually technology has this bias where the adopters tend to be younger. I'm just not sure this time. Yeah, I'm not sure. I'm not sure that actually it's age correlated, especially inside of companies, the number of stories. And now this comes to the corporate adoption point that I was, I was curious about second, but I am actually seeing. That the people who are like the best executives, and they tend to be a little old, like they tend to be a little older, they're not 24, are the ones who are adopting it the fastest. And they are having trouble getting their teams to adopt quickly. And it's very unusual that this is like how these things would play out, right? Especially for the non-technical teams. I'm seeing executives like Your CEO story is actually exactly on point to this point. It's like you're seeing the more senior people be more hungry to learn and roll their sleeves up really fast. And it's just not obvious to me that that's happening in the more junior roles that tends to be aged lower. I'm curious if you're seeing that. I agree with your point that it can be across the spectrum for sure. And there's been a, you know, a very young cohort that's been very much driving some of the cutting edge AI. I mean, young meaning 19, 20, I mean, a little bit younger than what we saw even in the SaaS days. And then you have a senior cohort that is great managers. Managers learned how to use AI earlier. They know how to do leverage. They know how to ask, you know, something to do something for them. You know, they're not necessarily individual tasks. And so I think we've seen a spectrum. But I do think in general there is a group that just hasn't latched on or found the magic moment or chose early to kind of fight it and say it doesn't actually help. I just met with a very strong engineer just the last couple of weeks working at one of the large, great companies out there. Publicly traded and, you know, not there anymore. But I, when I asked about just using the new code editing tools, I sensed that there was a pushback that surprised me. And I was like, and feedback was, hey, look, I would dig into that. I know you've just been doing it your way and you've had an incredible career, but I would, instead of saying you just use it as like a light review, I would really embrace it. Because what I'm seeing is that it's fundamentally different. And just last night that individual pinged me and said, hey, it was like 2 weeks later, Uh, I went deeper on what you said and I'm genuinely impressed and surprised by what I'm seeing. And so it's a small example of if people don't do that, they're not getting hired. And I think the last point is the managers getting promoted over the next 24 months are not non-AI natives. There's just no way the person that didn't lean into AI is gonna be the one that gets promoted by the CEOs, et cetera. So what will happen is we'll see a very fast acceleration of people that have really embraced AI and learned to use it the right way. Are only gonna wanna surround themselves with high-agency AI-native people. And once that happens, you know, like it's a different world and, and it accelerates at such a faster clip that I'm worried it really threatens people that are out of it. This is just moving faster than other innovations we've seen. And it makes me sad because nobody wants your friends or others to be outside of this. It's too good. It's too easy actually, once you get into it, you just need someone to help get you in it. I, I said exactly what you just said. I was at a reporter dinner yesterday, so there were folks from like Bloomberg, TechCrunch, whatever. And I made exactly this point about the AI natives will not want to work with the non-AI natives, and there's a big sort happening. And most of the table did not understand that. I think what you're— I mean, I understand it, trust me. I, I am living it. I, I deeply understand it. I am as worried as you are. I don't even think most people have fully internalized that yet. And that's my big yes and, and that makes it even scarier. You know what I think is the closest comparison to, you know, the AI native people not wanting to work with the non-AI native people is when you meet people who say you can't about anything that you know that you can do. You know, I'm sure you've met people like this that are like, oh, you can't do that, or that's not possible, or this person can't do that. And you hear the conversation, you're like, oh, I know this isn't true. Why have you got stuck on this? The idea that like everything is a red light and everything isn't possible. And I think it's, you just, you're sort of living in two fundamentally different realities. One is super pessimistic and negative, right? It feels bad to be around. And one is more optimistic, right? And, uh, yeah, more of a green light than a red light. The speed, you find that the AI natives go fast, right? They just see things as like, well, I can quickly just spin these things up and get a first rev or a detailed view of it faster. So there's a speed aspect. The second part is, and, and we're seeing changes, but when you see somebody come to you that did not use AI for some deep thinking on something, so maybe they're building out a new commission plan or a new way of designing your self-service pricing, and you see that they just sat there and did it all by themself. Clearly we want the human side, and I think we're all seeing it right now. We can't lose human thinking. Like, it is so valuable. But you need to also then leverage AI to see what the gaps were. What did we miss? What other new external thinking could I mix in? And when you see a human do both, which is, hey guys, here's a new messaging strategy. And, you know, I've run this through Adapt, which knows our entire business incredibly well and knows the external world incredibly well. And here's what it thought about my new messaging and how it rated it and what it also recommended. Well, then we're aligned. You basically are superhuman. You are incredible for who you are. You've used the best AI and you've used all the company knowledge, which I want. I want the, I want everybody's mind involved. But when you just send me a Google Doc of your thoughts, you're not AI native. I, I need to then go ask the AI. And again, I don't want the AI to do the thinking for us all the way. I want it to review, identify, help make us better. On the point of people being resistant, I wonder how much that comes down to how worried are you about it taking your job? And I, I think that when you look at senior leadership in companies, You're not actually worried about that at all because your job is not delivery-oriented. It's strategy-oriented. It's focused on the— you are used to saying, here's the thing that needs done. I'm going to go give that to somebody else to go do. If you are a software engineer who has been a software engineer for 30 years and your job has always been shipping code, it is a bigger threat to you than it is the, uh, the, the C-suite person who is used to delegating. And I think that's part of the problem here. I also think what Jack said is probably part of the problem too, but ultimately I think that makes sense. And there's your natural it's the fight or flight, right? Like as soon as there's something that threatens you, that's, that is our natural reaction. And so I think that's a lot of what we see here is like, if you're closer to the delivery of it, of the actual like end product, I think that there's more risk to you ultimately on these things. And, and I think that's valid actually. But the irony is that those who aren't willing to try end up being the ones most at risk. Can I, can I make a a provocation and then ask a question on that point. Okay. My EA directly told me, because we do this, like we're doing this like big AI challenge. I think I've shared this story before, but my EA directly told me, I am going to make sure that 100% of my job is done by Claude. And nothing has made me want to keep her more than when she stated that objective. Obviously, right? Like every, this is like a CEO's dream that that's what someone would come to them and say. Sort of. We'll come back to it. Okay. Okay. So, because, because now I'm sitting there thinking she has high initiative as she's automating these things, she's showing me that she's thinking about the outcome versus just arbitrary delegation. And now I'm like, oh my God, I can trust her with all of these other things that I previously would not have even thought to trust her with. Right? So she's showing me the skills and attributes that I, that I care most about in, in terms of more tasks and other projects and things like that. So at least for me, that is like my view of that. The question that that begs is, first of all, and I'm curious, Jim, it sounds like you have a provocation on this, but is that generalizable? Right? Because I don't know if it's generalizable and perhaps we will say it's not generalizable globally. The other thing I wonder though is how much of this is just what is around you? So like, I, I've recently been asking my team to come spend some time in San Francisco, and I think it's because if you walk around in San Francisco, it's really hard to not become AI native. Real— like, I can't see a billboard that is not something AI oriented, like a cab, a bus, a train. Like, I'm like, no matter what I do, I cannot escape it. And so, um, I'm wondering, and, and so it, there's like this cultural question that I'm very curious about. It's like, I can tell that this city, something special is happening right now, but I am wondering inside of organizations, what is the cultural setup? And Jeremiah, one thing you are saying is people are worried about whether or not their job is going to be taken, but I think that begets a much harder question of what is the culture in your company as to whether or not that person is afraid that their job will be taken, right? I mean, I think I've, I've two, two thoughts on that. I'll start on Jeremiah's, 'cause it, you know, talking about like the coder who's worried that, you know, suddenly Claude Code can take their job. Clearly the way that we do our work has changed, like on that role. Absolutely. But you're also seeing challenge right now where people are hiring engineers that don't actually fundamentally know maybe what great code looks like. And so there's a lot of vibe coding. It's creating some friction and some thrash. And let me give you an example. Take the thing that we each know really well. Like, what is the thing that you are literally an expert at after all these years of experience? When you ask AI on one shot to build that thing that you're an expert on, it does not get it right because you know, you know what the right answer is and you're like, whoa, whoa, whoa, this is wrong. But you know, this is good, but change this, change that. And you'll do a few revs basically teaching AI and it'll be like, yes, you're right. Yeah, I need to do it better. But when you do things that you don't know very well, AI looks flawlessly perfect. So when you take the coder that doesn't know coding that well, then the vibe code is perfect, right? What, what are we gonna do with it? And so my point in that is that as you see people that are really strong technically, And then they learn to use these tools in incredibly special ways. I think they have a very large unfair advantage, um, over everybody else right now and can do magic with these tools that I don't think the tools are right now doing on their own. So that's one, one point of it. The second quick point, um, on the EA using 100% Claude, completely love AI native, but there is like a, a balancing act of the human. Like we want ultimately the humans to win, and I know that if you're EA, use Claude to write the most perfect authen— you know, AI response to me when I email you. I'm now over it. I don't want it. It's not— it's, it's, it was cute a month ago. It was cute maybe 3 months ago. I want an actual human response. Now, if Claude helped her to, uh, you know, an AI enabled her to be fast, to have some data points, to do something really authentic, it's great. But I think we're all starting to realize that human communication, there's a lot of trust built in the words we use. And there's a lot of noise that's coming. And so we wanna use the deep thinking, you know, that this company brain to help people think, but we still want them, you know, critically putting pieces together, putting their own stamp and touch on it. And I think people are begging for that right now. Okay. I definitely agree with you. So I will amend my statement. The things that she are, she is doing is actually all of the EQ parts of the job. Um, she is doing herself because that's what she's exceptional at. Right. And everything else, but like the task organization, Rish, like the annoying me to remember to, for that way I actually do what I'm supposed to do. Like, have you done these things? Here's what you promised people. You've not done it for 4 days. All of that is all automated. And, and so like she has struck this like exactly the balance that you're describing, which I think is critical and, and totally underappreciated. Uh, Jack, I don't know if you remember, but we made this joke in a previous episode around like humans should just worry about like the interaction with other humans and the like human level thinking. But there's a lot of thinking that computers are better at and, and, and computers should do the computer thinking. So yeah. Yeah, definitely. I think the EQ is a great way to think about that is that people really do need the EQ. I think if you talk to friends looking for jobs right now, what's challenging for them is that everyone's looking the same. They have a perfect cover letter, they got this whole presentation, and as a I can't, I don't know what it is. It's like there's a middleman that literally delivered their career to me. I need to see how they write and what they do. And I want them using AI, but not necessarily to be synthetic on their voice. And so even I've been training AI, don't, don't write, if you're going to try to take a stab at something, don't write something fake. Like, oh, I loved your XYZ. Give it to me straight. You signed up. I noticed this. I want to do this. Let's, let's be really functional about this., but the IQ side, the ability to do special powers, an EA plus AI should be able to do so much more and see a bigger thing and reason differently. And that's the magic. And to understand your workload. I mean, if you have a company brain and she had access to these tools that you're in, the speed in which she could react would be dramatically faster than waiting to connect with you. And the ability to build triggers of like, oh, if this meeting happened and you looked at the transcript and here's the notes, now use your EQ to have the perfect follow-up or gift, et cetera. It's magical. We just need the human touch still. Yeah, I totally agree. Jim, I'm really curious. Another thing that I think a lot of software founders are thinking about right now is what are moats that will persist? And so I'm curious how you think about that in the context of Adapt. Um, and then how do you think about the— I, I get very curious right now about how to think about capitalization. And like what are outcomes that are possible? Because the mark, the shape of the market has never been stranger, of the capital markets has never been stranger. And so yeah, I'm, I have two totally unrelated questions, but I am very curious on both. Uh, yeah. How do you think about moats and defensibility? And then totally separately, how do you think about what is the right way to capitalize a business like yours? Yeah, it's, it's, it's And what a, I mean, how lucky are all of us to be in this right now? I mean, honestly, I've seen different waves, but just think about what we're doing right now. We are just at an epic, epic time of innovation and nobody knows where it's going. I talk to every friend, the smartest people that are in it, and you say, where's it going? You talk to friends from OpenAI, et cetera. There is a sense of, we're not even sure what that next OpenClaw moment is in 3 weeks that might happen. Like there's this really fun early stage that we're in. I saw it back in the '99 era when we were doing, you know, Evite.com and those startups. But this one's different and it is moving at a different speed because it's just, it just, there's so much more capability, um, in it. And so when I think about, when I think about the moats, when I think about the space, one, how neat to know that you're working on something that is going to be an inevitability. Every single company will have a company brain, not a bunch of AI tools, you know, not a license where, hey everybody, here's like an individual AI tool that you can use to do your work. That was neat. That was like a 2025 era. No, hey, we run, I think the first interview question you will ask in about 2 years is, hey, before I even go, what's the AI? What is the company brain AI that we're using? Because I, I've been working at Adapt companies and it has been the most insane experience, but I've heard friends are at these other companies running on the XYZ and it's miserable. It's a surveillance state. It's counting mouse movements. It's, it's some of the stuff we just heard about with Meta doing, right? Where they're, They're training off of this. And these are decisions we need to make. And I talked to our team about this. Like, I really want us to be on the side of making it so that our friends and family, when they go to work, don't say, you just screwed up work. I mean, it used to be great. Now all of a sudden, no, I want them to say we literally threw on Adapt or I joined an Adapt company and I've never had so much fun. And the way that you've sort of non-emotionally brought our teams together, the way you sort of review analytics in a way that has no politics, the way that you've just augmented everybody. And so the first part of that is we're playing in just the most massive market. I mean, there is trillions of dollars of transformation. Every single company is gonna have a single AI brain. Um, and I've just watched SaaS companies chase market share. It, it goes to many different flavors on it. And I think this will be a horizontal play. I think there's a lot of vertical AI. They've done quite well right now. They'll continue to have, uh, good growth, but this one's different. Once you start to put in the horizontal layer and what we see, and Jeremiah, you, I'm curious, you start to realize, I'm not quite sure I need this other thing that just ties into this one silo of, of marketing or finance. And over the next, you know, 12 to 18 months, you're gonna see this like virus, this productivity virus of, hey, why don't we connect more tools to the brain? The brain just got smarter. Let's have more knowledge in it. How do we create a reasoning engine that's always on and persistent behind the scenes, which is a whole new way to think about AI than just ask, query, and generate. No, we're about to move, and you know, this is stuff we're working on into a far more proactive, always-on system of intelligence that truly starts to understand patterns that the teams would not be able to spot on their own, and they're not even asking for. And so the moat that— these are the moats is removing from a tool. You ask, you get an answer. You know, at Adapt, what do you think about this? I love it. That's what we do. And it's powerful. At what point does Adapt see the bigger picture? Bigger than a CEO could see. Start to cross-reference patterns of why did we lose that deal? That's not just a query right now where you go to HubSpot and a few other pieces. It's already identified a product decision 6 months ago that had a chain reaction, and it's identifying and bringing this to us. That's what we're about to see. And those engines, that reasoning engine, is going to be different moats that are different than just the frontier models. And how we apply the frontier models to these extra reasoning, um, will be a big deal. But if you're model agnostic and tool agnostic, then you really have the ability to tap into all the best models. So as you know, ChatGPT Image 2 comes out and you're like, that's amazing. Boom, pop it in. And so I think that the mode is being agnostic. We know that the enterprise likes flexibility, likes to be able to make bets and make sure that it'll always be the most intelligent enterprise. And then two, it's what we do with this that goes far beyond ask and get a quick answer back that I think is going to be the next iteration that we see. I mean, we'll be launching this at the end of the month. It's a very powerful extension of what we're doing. I think one of the things I appreciate as a user is Adapt is the, the first, uh, it's actually like a really good response half the time, the first response, right? And then as, uh, when you're asking something new that you haven't done before, and then the other half of the time there's something missing. There's missing context. There's, there's information that needed, that Adapt didn't have to be able to give the right answer. But what I love is you basically just look at me like, okay, well that's not quite right. Hey Adapt, can you add this piece of context? And then you save it and it's done. And then every time after that, it has that context and it's, it's getting better. And I think that's one of the things that's going to be really interesting to see. I mean, honestly, Jim, I wasn't thinking about it. I am not thinking about this problem the same way you are. So just to even hear what you were describing there, I think is really cool. Like I see the— I see that vision, I see what that can be. And I think it's really interesting. And one of the things I would love to talk about, I love talking about pricing. And I think this is a really interesting thing when it comes to not just Adapt, but any sort of AI tool where it's token-based, basically, like how you think about that. And just like, to be fully honest, as a user, I mean, our spend is about 4 times as much right now as I thought it was going to be when we signed up. And that's not because it's like you're charging me more. It's because it just keeps compounding, right? The more people start using it, the more like it just keeps going and going and going. But there's an upper limit. There's a day in which I'm going to wake up and be like, Wow, this is such a massive line item. Can I justify this? And I think the— this is something I'm thinking about with AI in general is what, what does that look like? I mean, Claude, for example, just got way more expensive. There's a world in which like token costs could rise to the point where we start to look at things and say, I'd rather just have a human do that, or I'm not going to do that at all because it's just too expensive. So I'm, I'm curious how you're thinking about that and the future. And obviously if you're the brain, there's a premium that you can charge there that couldn't charge if you are in a silo. Um, but I just think this is a really interesting, um, piece of, of the puzzle. I think this is gonna be the hot topic. You know, we are in a 6-month window where, you know, AI was very inexpensive, right? We were paying $20 per employee or discounted on maybe an enterprise plan. And you'd always joke like, if, if the average employee, uh, pick round numbers, you know, $10,000, $20,000 a month, just using a ballpark, you know, examples. And then to give them $20 of intelligence,, we all knew that was very undervalued for what we were seeing, at least if it was being used. Claude Code and Cursor showed you when AI actually creates value and does and increases productivity dramatically by doing work, you will pay far more for it. And with very little friction on that, I don't, I still don't think CEOs are pushing heavily back on that. They're concerned, they're consuming budget way faster, but I don't think they're willing to say, you know what, I think that was a neat experiment. Let's go back. And the second, and this will get into the trend, we can all just use less smart AI if we want. I mean, we could make this very inexpensive and have a high schooler do your work for you instead of a PhD. And I kind of kid when I say that, but it's like, boy, to use the absolute best AI on these things and know that we're still, I don't know. I mean, what's your average per, maybe I'll use general numbers. Let's say your average per employee went from $20 a month. $200 to even $500 or $250 or $500 if you averaged it across your team. That's about where we are. Yeah. Think about how small that is. I mean, could you pay anybody on this planet to do this kind of work for $500 a month and what time it would take? I think that we're kidding ourselves when we think it's expensive. I do think there is a limit and there's gonna be a huge focus on efficiencies and new ways to do pricing. So I, I hear you on that and that's something we'll be thinking about. But I think we're still in an era that if it's used thoughtfully, so you're gonna see heavier inspection, heavier accountability on ROI, which will be very easy. Even just ask. I mean, I could just ask Adapt for your account, build the ROI case study and really help to make sure Jeremiah realizes the value we're creating and it'll have it in an hour. And that's— I'm not, I'm not questioning the value. There's a reason we're spending 4 times as much as I initially budgeted. It has value. But actually one of the things I think is really interesting, like when in the things that we're building for our customers, I'm thinking about how where we sit, we can actually help reduce their token costs in Cloud or whatever they're going to use to connect to this. By being smarter with the way we pull the information for them and that kind of stuff. And I think if you're the brain, you actually have a massive ability there to where like over time there's probably a world in which you can actually deliver better outcomes for less tokens than you, than you would be able to get with just connecting Claude or whatever directly into a system because you actually have more context and more ability to— That's exactly right. To tune those, the inputs to get the right outputs. So I, That's, that's how I think about it and what I'm hoping that ends up becoming ultimately. But I don't know if that's part of how you're seeing it too. It's exactly right that as you start to have all that context and restoring, you know, this context and doing things with it, we don't have to do all these tool calls every single time. I mean, there's a different way to pull this context and share that. In general, it's an ROI game. How much value are we creating? Yeah, we can tune these things, right? I mean, it's all a tuning game of how much, how much of the good stuff to put in. But I think that most people are pushing us towards keep it going. Like, and we've prided ourselves, we always kid ourselves by saying we want to have the single best answer. Like, we won't price optimize right now. Uh, we don't even give you an option to say use the dumb model. No, we want to actually give you the single best models we can find. Now there's, there's work you can do on, on, on tuning this, but I think the world still is on a zone of actually give me the best answer possible for this and then help me maybe understand if there's use cases that we could maybe not use AI as aggressively on, but Uh, I think we're still in the zone of lean in and unlock and accelerate how people achieve their missions because it's just, it's been too slow and we're in a massive productivity acceleration and most CEOs I talk to don't want to slow it down. And I would also say the best still isn't good enough actually in a lot of cases. So you may as well just keep pushing that until you get to the point where it is good enough and then figure out how do you make it more efficient, more cost effective, whatever that is. There are people using Claude Code for free right now, uh, because Claude Code leaked their entire thing online and there's now like a repo that's like super popular of people using Claude Code for free. And then there's all the Chinese models. Like these things are becoming so cheap, so rapidly. There's going to be people, you know, there's no coercion. They're choosing to use the very best models, always up to date because they want the best result. But man, like the AI we're using now is going to feel like the AI that was available in like November. Last year in just a few months, and we're not going to want to use it. You know, we're blown away by it now, but like in a couple months' time, we're going to be like, oh, I can't imagine using anything apart from Mythos, right? I'm only going to use Mythos from now on. Right. Um, so I don't know. Someone put you on Sonnet or put you on a, you know, 5.4 when you knew that there was this magical thing that literally cost, what, $100 more a month per— I mean, when you actually think about how much we spend in these companies, I don't know if this actually is that big, especially if you look at the productivity gain. There'll be areas that need to be more efficient, and this will be very easy for the industry to solve. It'll just start to build more audits, and actually we'll just use AI to solve it. Hey, Adapt, is this a good use? Is this a good ROI positive query? How would you think about this? And there's a lot of ways to just use AI to manage AI. One of the things that we spoke about earlier was driving adoption in teams or even people, you know, outside. One of the awesome things that Adapt is doing is bringing people who are not AI native, could be AI resistant for whatever reason. There's, you know, thousands of reasons, but making it So that they can feel what we all felt when, I'm not sure what tool it was for you. I feel like the tool changes fairly often for me, but right now it's just Clore Code in Terminal. But like first it was Codex and it's this amazing feeling when you're like, okay, all these complicated things I used to have different specialists do the sequence of baton passes and it would take a week or longer to get done. Suddenly I can get done every day in the background and not think about them anymore. How incredible is that? I mean, yeah, I wonder what you guys think about, gamification and bringing that experience to your teams? I've heard some people talk about like token maxing, right? At their teams where they're saying, hey, use as much tokens as possible. Yeah. How do you guys think about that? One of the use cases that I think just brings a lot of joy to people is if you connect all the tools you use. So let's say you use a HubSpot or Linear, all these, all these different platforms using, we have a, we have a product called Adapt Apps, which is like a cursor cloud code type experience. But when you let just like, we have an AE on the team. That decided to just in natural language build a new type of way to manage his accounts when he gets on a call. So instead of logging into HubSpot, which we don't want to vibe code HubSpot, that, that would not be a good use of time. I wouldn't, I don't, I don't think any systems of record should be vibe coded right now, but to add a new UX layer that is very visual, that has trend charts of usage, that has the billing data from Stripe, that has the usage data from BigQuery so we can see who's using it and is like a really full functional app and then has Adapt Intelligence of like signals. We never log into HubSpot. It writes back. And that's the example where, Jack, I think that when people start to build their first apps, they have this like huge smile. They never thought they could do this. They in the past would have to have an engineering team working with them. And then when they can share it, which is what Adapt's about, it's team-powered AI. This isn't the, this isn't like a cloud cowork where you're just very siloed. Um, you share it. So then someone else uses the app to do their work or edits it. It's a pretty insane experience of just how it changes how you work. You basically have the most perfect UX. For any situation you want. We have a TV on the wall, like a frame TV in the office. That's just an Adapt app that visualizes all of our usage in a funnel flow. I mean, literally in the last 24 hours, all the messages, connections, users, bubbles are floating. It makes noises and it's just someone built an Adapt app and said this would be a more fun way to manage our accounts. Yeah, I've started using those just to build, to prototype things for customers. I've got the information I need connected to be able to pull their information, build these things out. And, uh, I, I think hopefully this comes for you guys at some point, like actually being able to share those externally would be cool. But even for now, just being able to take screenshots and say like, hey, what do you think about this? It's something I can just do from Slack over the weekend. Um, I'll literally like, hey, I, I don't wanna log into Adapt at all. I'll just ask it to take a screenshot and send it to me. So Adapt is literally taking a screenshot of the app that it's building, and then I, I'll look at it and I'm like, okay, cool. Like edit these things. Well, what, what it's become for me, cuz we actually got started with Adapt around the same time. That all the OpenClaw stuff was really going viral. And I actually started playing around with OpenClaw and I was like, wait a minute, this thing's actually better for me because it's embedded into the business in a, in a way that I would never be comfortable trying to embed OpenClaw because of the, just the, the risks of that. And it's, I don't think it's there. Like there's, it, it doesn't have the same level of flexibility, but I actually don't want that. I want a little bit more control. I want things to be done in such a way that is shareable across the team and all that kind of stuff. And, and so it doesn't solve all the same problems, but I do kind of see this brain concept that— or computer, however you want to describe it— that sits on top of your business. It definitely— that is the future as far as I'm concerned. So for me, it actually— I actually stopped. I threw away the whole idea of using Open Cloud because it's actually not that useful for me in a business context. And this did that for me. Hey, Jim, just one closing question before we wrap here. If I'm trying to figure out what is the thing that I need to do in order to get my organization to become more AI-native, like not, not really tool-based, but more culture shift,, right? Because you guys are providing a tool that can help. Um, and, but there's a, there's a big other part, which is just the change management part, right? What is your top 1 or 2 things that you would suggest to like a CEO or a CIO, whoever is responsible for this? And it's usually the CEO is responsible for this. Like, what is the thing that you are telling them? Like, hey, I know you're like, you feel like you're banging your head against the wall trying to get your team to be AI native. Here's like the top 2 things that we see across all the companies we work with. It's an interesting— I'm thinking through how to answer that thoughtfully because when I hear that, I think like AI empowered, which is, oh, we want our, you know, if we don't talk about the tool, then what you are talking about is I just want people to use more AI to augment their thinking, to maybe help them see it differently. But when it's all just done in silos and not connected, I'm not, it's hard to use the word AI native with that because all that really is is I use an AI tool to do things in like a weird, not weird, but just in a very isolated way. And I think the key— Sorry, I'm saying like the choice, I mean, the choice of a tool happens in the course of doing that. But at some point, someone has to have the desire to want to work in this new way, right? That's, I guess, what I'm talking about. I think that it starts at the— somebody has to have the initiative. In our world, it usually starts with VPs or a CEO. I'll use an example. Let's say you created a skill in Adapt that was like the most perfect way that you like to connect with people and build relationships for your EA. So you save the skill and then your EA starts. You said, oh, by the way, when you— every time I have a meeting, can you just run that skill or just even ask Adapt like, how I think about this, and it does it. You've now just made this person far more AI native. You've codified a way of thinking and taught them that they probably could codify a lot of their thinking. So if they hire a second EA or expand the team, it goes. And so I guess it comes down to how do you create these magical moments that others see that expands it? And maybe the last point is Slack is a wonderful way. We use, we use a concept of being instantly AI native because as Jeremiah mentioned, he, he took the time to connect APIs into his account, but it was done at the organization level. So as his team jumped into Slack, all they had to do, and he just probably sent a message, is just say, ask Adapt what it thinks. Say, @Adapt, how would you do this? Or @Adapt, this customer has this problem, what should we do? And suddenly they had an aha moment that everyone can see in a team-powered AI. And so I think it's about, the reason why I push a little bit is without the connection to tools, we're just in a little bit of like AI, consumer AI in the business world, which we've seen in past iterations. But when you connect it to the actual work we do and how we do it, it's like, it's like an experience that no one's had and that you couldn't ever imagine not having. So like if someone said, you know what, tomorrow we're gonna turn off this connected brain experience and you're just gonna have your team in Slack and your team by itself doing the work, it would feel like turning the Wi-Fi off. You still have your laptops, we're still all together, but we're really like hampered, like 90% down. I think that's what a connected intelligence feels like when it's connected to the tools. And it's very different than everybody having a ChatGPT license where if that went down for the day, I think most people are fine doing their job for the day. It's painful, but they, they would, you know, the work would go on because at the end of the day they were doing the work and they were simply asking another tool to whisper to them a better version of the email or to create something that they then as middleware, like a human middleware, would go add to other systems. And that's what we're fundamentally changing. We're getting rid of the human middleware and just letting the humans do the part they're so good at, like the EQ thing that you mentioned. How many people do you guys know that are passionate about Google Docs? Or Gmail sending emails? None. Right. Demand drives supply. I felt like you gotta like ask for it right from the team. That's right. Yeah. I gotta jump here too, but Jim, can you just kind of give everybody like a little last thing on, uh, Adapt and how they can find you and, um, where to go if they're interested in it? Yeah. Um, so adapt.com, company brain powered by the Adapt AI computer, and you can now just go to adapt.com, sign up, and if you use a business domain, you get $300 of credits. Our team will reach out as well. And we will help make your company AI native. Yep. Awesome. And, uh, tell when you found us here too, cuz I think, uh, it'd be fun to, to see if anybody decides that they're gonna go use Adapt. Uh, yeah, we gotta set up a code or something. Love that. That's great. Yeah. Super fun. Jim, this was great. Um, I, I love hearing these stories. You're just, you're, you're out there doing it and, uh, helping lots of teams become AI native.