In early 2025, Mitchell Jones sat in his basement in Ohio and burned a few hundred dollars trying to build an AI agent that fills out forms, a job that should have taken an afternoon. He kept paying for access to the same models through different wrappers. That friction turned into a company. Bill talks with Mitchell, founder and CEO of Lava, about how that basement experiment became a business, and then a different business. Lava started as a gateway: one key and one bill for every model, on the bet that AI is a utility and every utility ends up on a meter. Then users at an event kept asking to plug it into Claude. Instead of telling them they were using it wrong, Mitchell asked why. That answer turned Lava into what it is today, an AI agent that works inside the tools a business already runs on, like QuickBooks, Gmail, Salesforce and HubSpot, with no engineering required. They get into why Cursor's simple layout (your work on the left, an agent on the right) beat the blank chat box, why customers rarely describe their pain cleanly, and why you should fall in love with the mice problem instead of your mousetrap. Mitchell makes the case for planning in three-, six- and nine-month windows while still building an all-weather business, and for building something that gets better as the models get better instead of fighting the frontier labs. Bill closes with a challenge for leadership teams: if you aren't using AI every hour of the workday, ask your AI what you're missing. Then find the enthusiasts already in your company, buy them lunch every week, and have each one share something cool they did with AI and something they're stuck on. Before Lava, Mitchell co-founded LendTable, a Y Combinator company he scaled to about 40 people and $40 million raised. He's a Forbes 30 Under 30 honoree and has worked at Meta and spent time at Yale. In This Episode How a basement side project turned into Lava Why AI is a utility, and why every utility ends up on a meter The customer question that turned a payment gateway into an AI agent What Cursor got right that most AI tools still miss Love the mice problem, not your mousetrap Why founders should plan in three-, six- and nine-month windows right now Bill's hourly AI rule and the weekly lunch-and-learn for your flag bearers Why you shouldn't trust an AI answer that came back in two seconds Guest Links Lava: lava.so Host Links Scaling Coach: scalingcoach.com Busy Is Broken: busyisbroken.com CEO Behavior Study: scalingcoach.com/study Scaling Up Business Podcast with Bill Gallagher
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
It's been a long time since I've been in the studio. Almost two months since we've recorded a show. And it's nice to get back. I feel a little rusty, though. What do you care about that? You care about my guest. He spent... Just a few hundred dollars trying to build a form filling agent should have taken him afternoon. He kept rebuying access to the same models, different wrappers. That friction became a company. We're going to get into all that and more and how the initial thing evolves into something else and actually becomes
interesting. That's really our story today. It's very AI related. Listen in. You know, I'm playing with new music, but then I listen to this old music. I'm like, oh, I kind of like this. I haven't... It's a nice, I know. I like it. It's a, it's a, it's a good intro music. You know, you got something to head nod to. It's, it's, it's, if it ain't broken, don't fix it. Bill Gallagher, Scaling Coach, host of this super groovy Scaling Up Business podcast. Our show comes every week,
well over seven hundred episodes now, ten years in the making. And we love talking to entrepreneurs, gurus, authors, everybody to do with scaling your business and all the trials and tribulations that go into that. Look for the show and more wherever you get your podcasts, like, subscribe, notifications, all that kind of groovy stuff. I got a new book out. Busy is Broken. Busy is Broken. Where's the book? Here. Here it is. If you're watching on video, there it is on screen. It's out now. Get it wherever you get your books. It talks all about how you become the bottleneck even after you built a team and
how to break the habit and scale the company with even less effort. Busy is Broken. Busyisbroken.com, scalingcoach.com, wherever you get your shows, all that good stuff. All right. We're joined now by Mitchell Jones. So Mitchell is founder and CEO of Lava, based in New York. And the company does all kinds of work. It's evolved. He'll talk about that. But evolved into something that helps you integrate with AI with your existing software. So you got existing stuff.
How do we infuse AI with that? Everything to do with all that technology. A lot of old ways of doing stuff are hard work. They're focused on that. He's going to talk to you about how it came across and all that. Mitchell, welcome to the show. Hello, hello, Bill. Thank you so much for having me. Super excited to get into it today. Yeah, so among other things, Forbes is already under thirty. He co-founded a company called LendTable, a Y Combinator business. He worked at Meta. He spent some time at Yale. So lots of good things.
And I'm delighted to have you on the show. It's a pleasure. It's a pleasure. And that's all a bunch of old stuff. So, you know, thanks for, I appreciate you highlighting it, but it's all the old stuff. Well, let's get into the new stuff. But before we do that, let's talk about how you burned a few hundred bucks on an agent who was supposed to do some things. And then what did you see there? Tell us that story. yeah so you know I think you're talking about the original founding of lava and uh
you know it's funny most good startups don't start out as this complete idea with market research and data and all these things it usually starts out with you were playing around in your basement and you ran into some issues and you just were curious, like, why does it work this way? And so for me, you know, I just finished up working on LendTable, which, you know, I'd worked on for about five years, scaled it to about, you know, forty employees, raised about forty million dollars for that business. And then, you know, I was just starting my next one and,
you know, was back home in Ohio for a little bit, had moved back from San Francisco. And all I was trying to do was this is early twenty twenty five, just so that everybody knows what it was. The big thing that had changed since I had started founding LendTable, which was back in twenty twenty, was AI had taken off and I felt like I had been on the sidelines of that. I didn't really understand the depths of what it could do and what it couldn't do. So naturally, the way that most people who like to build things do things, they play around with new technology. And so I was just sitting in my basement and I was like, okay, well,
everyone says you can build so much stuff with AI. What's something I do all the time that I would love to just see if it can do a project on? And I was like, I fill out forms all the time. And so imagine if you just make an agent or it wasn't even called agents back then. It was like, imagine if you just made some software that would just be able to fill out. I'd give it a form, finds a field, fills them out for me. I wasn't trying to sell it or anything like that. But in doing that, I just recognized how powerful models I thought were going to become. And, you know, I didn't understand the shape.
And, you know, I think that probably the moral of the story today is like how much the journey of a startup changes over time. But back in early twenty twenty five, there was a couple of truths I saw. Number one, I didn't know that we would be building lava.so as it is exactly right now, which is we make it really easy for the average business, the normal mom and pop shop, the large business to integrate AI agents into your product. You don't have to use new tools. You don't have to have engineering required. We make it easy for you to have a real assistant that works in your
existing tools and right alongside the existing software and apps you already use. But it didn't start there, right? The first thing we saw was like, wait a minute, These agents are really, really, really powerful, but we think that these models are our utility. And if models are our utility, every single, utility business I've ever seen in my life was priced one way. It was priced on a meter. Your water is a utility, it's on a meter. Your electricity is a utility, it's on a meter, right? Cloud, although you think about it like a SaaS business, cloud is utility and it's on a meter.
And so back then we were really setting out to say, how do we make it really easy for you to route you route model spend, route compute, and also if you want to be able to essentially have one source with which you're paying for all of your compute. So back then people were, oh, I need to get an API key from OpenAI, I need to get an API key from Claude, I need to get an API key for Midjourney. And we said, actually, you can get one key from us and you can pay us for all of that.
That's where we started out. And, you know, it's been a huge change since then, right? That's January of twenty twenty five when we started tinkering on that idea. And so the moral really here is like we started with a bunch of stuff there and we just kept editing. And that's only ten months ago. I know. But in startup years, that's like forever ago. That's the thing that's so, so exciting. You know, twenty twenty five was, you know, January twenty twenty five. That feels like that was so long ago. Right. And although it's only been a year and
some change, you just see how quickly ideas change and how much this market grows. And like, you know, right now, AI is moving so fast it can be dizzying for a lot of founders, for a lot of people. But you just have to be brick by brick. You think about early GPT interactions and what that was and then what it was like a year ago and what it's like today and that kind of thing and the complexity of what you can do. And I think like the last year I was using it for a lot of writing and creative work and as discrete
elements. Today, all of those things are working in combination. We're building, we're coding, we're doing multifaceted, multi-step chain scheduled things. We're referencing existing assets and going back to things. We've got live real-time things hooked out to customers and clients. That's a really different world, right? Completely. And, you know, I think we've even seen that, right? Like when we started building lava,
I think the best model out at the time was like Sonnet four, four, four, four, five. It was something around there. Right. And, you know, back then, you know, all of our core infrastructure was built by hand. Now, if an engineer is building entire core infra by hand, it's like they should be planning it themselves and then coding almost always with a coding agent. It just feels like that was so long ago. But I think you said another thing that I think is really powerful, and it's really what we focus on at Lava every day, which is how do you make it so
that the average day to day business, whether it's a software company in San Francisco or someone who's running their own SMB, coffee shop, how do you make it so that they can actually use agents in a production level in their actual day-to-day work? Whether you're trying to just analyze your spend with QuickBooks, whether you're trying to categorize your ERP. We have customers all across the board who, because we make it so easy to connect and add an agent into the tools they already use,
They feel like they have essentially just superpowers inside of these apps that are still kind of legacy. But because they're just using us inside of them, it feels very efficient. But, you know, I think that the real thing to highlight, you know, that I always love to say is like it feels so clear when you're on when you're at where you are now. But, you know, when we were starting a year and a half ago, we just had a thesis. Right. And, you know, a lot of things were wrong. We got some things right. We got some things wrong. But you have to just keep iterating. I think that was the thing we saw in the early journey.
Well, people do need gateway and they need console. I mean, that was kind of the thing with Cursor and led to their valuation and acquisition. That kind of thing is like, they're like, let's give you an environment where you can use, and we're gonna develop our own language model, but we are gonna just access any of them and use them fluidly or let you pick them or pick them for you and give you one environment where you can do your coding. And it was super, super popular.
And that was happening there. Now you're seeing that more and more. I think the other thing that now there is is the need to think about the safety checks and the gates. What are the limits on things? Because you'll build something and it'll seem like it's working and then all of a sudden it takes a left turn on you and you're like, Where did that come from? You could do that yesterday. Or why are you jumping ahead? Or wait, why did you send that email without my approval? Tell me about it. And like you said a couple of things,
which is like actually Cursor was a North Star of a lot of what we did. Well, number one, we built a gateway pretty early on. So that was actually the core thing we built. We made it easy for you to pay and route things. All a gateway means for those that maybe don't know the technical terms is it's just literally, it's a piece of software that allows the end user you to call any model without having to go set up relationships at every single model provider. A new model comes out every week. Do you want to have to change that
or do you want to just be able to say in natural language switch from Opus four six to GPT five point four? Right. And that's what model that's what gateways did back in the early days. What we found was similar to what cursor found for engineering is even gateways are still at a level on their own that sits still in the engineering part of the stack. What we try to do with Lava is be able to get businesses where they don't have to learn these new technical primitives. You know,
what the average business wants to do is they want to say, I use QuickBooks and Stripe and Ramp and whatever tools I use. I just want to connect those to an agent and let the agent log in and then be able to talk to it. And the thing that Cursor nailed in the early days, and you mentioned this too, like Bill said something that I think is really, really elegant and really important to highlight. You know, Cursor is now known for its model Composer, right? And it was called Composer two point five was their big one that got really, really popular.
But before that, it just started with two things. It was all of your code on the left. a little pane on the right that was your agent pane, and you could just choose whatever model you wanted to use, and you got to prove and verify, as you also talked about, before inputting it into your code. We do the exact same thing for every single product, whether new, old, or legacy, and those are things that we do all the time. So, yeah. But yeah, so I think like, you know,
really, Cursor really laid a great foundation there that I think a lot of other businesses sometimes miss. And like, if you think about every single, I'm going to automate your work right now, whether that's like, you know, I think DoorDash just made a bot, Spotify made a bot. Obviously there's a Grok bot, obviously there's coworker. All of those things start in a very specific place. It's very different than where Cursor started. They start with this like empty screen where you have to kind of imagine the
magic of what needs to happen. Whereas with Cursor, the magic of it was I'm just in my code. And so that's really a lot of what the inspiration of Lava took is like the approachability for the average person. Like why is the average knowledge worker still just chatting with ChatGPT is like, they're having to imagine what the possibility of what it can do. With Lava, you just see, oh, I'm looking at my Gmail or I'm at my Salesforce or at my QuickBooks. And there's just a little pane on the right that says, hey, what do you want to do here?
And it can do all those things that feels like magic. But I think you said something that's really, really was really important is like you brought up Cursor. And I think it was an awesome moment for not just AI, but truly for like, what is a good way? to integrate new systems, right? These AI systems with old stuff. And I think that they did a great job with it. Yeah, and with the teams that I coach, I see this again and again. There's definitely still people out there who are resisting AI for coding.
I'm like, you are just out of your mind if you're doing that. They are going to fail and your company is going to fail if that's prevalent. There's just no way. I've had two companies now fail because they were resistant to the AI adoption. They related it to as a thing we're going to think about next year or whatever. And when I watch even like, you know, my son's been involved, I've talked about it in many shows before. And the first versions of the image generation, you know, they were done by hand,
running on a huge ass machine in his garage. And, but, but operated with, you know, just in a terminal on his, on his laptop. And a lot of it done on the deck outside our house, right? Like, and building that, that became Midjourney Two, eventually through, and roughly two through five, I might be wrong about this, but six and seven entirely like machine coded. He stopped doing hand coding of anything.
So, And now it's agents running agents and testing and doing graph engineering against things like looking at how does this one, let's try that experiment and let's, and let's run some numbers on it. Okay. That did a little bit better. And let's, you know, so it's, you've got agents running agents and you're kind of supervising and you're conceiving of projects that are running across it. And then that raises the new issue, which comes up a lot in my workshops. Now it's like, wait a minute, do I want to give access to my
accounting, to my email? Like where are the guardrails on some of the things like that? And again, And who do I trust with that? Yeah, correct. Who do I trust with that? And like, you know, you said a couple of things, too, that I think are important for like the scaling journey as well. It's like, you know, people often think so much about the end of the destination and like you highlighted it really right. It's like the first version of everything is so scrappy. I remember the first version of Lava. We literally had to try to make the
routing work reliably. Back then, everything wasn't standardized in this way, where it's like, oh, if you want to route to some model, there's a base substring with a quantization level. All of that stuff wasn't really that high fidelity back then. You'd have to put these long, long lists of if it looks like this or if it's coded in this way. And that's gotten better and better and better. But I think another thing you said, too, which is one of the things that was very clear for us was we tried to
build a business where as the models got better, our business got better. Like that was just not the thing we thought we would be doing was ever being a model player because they have so much research budget behind them. The real goal is how do you think about where they're going? Right. Exactly. How do you think about where they're going and how do you try to say, imagine what my product can do right now. If I harness this well, it's going to be able to do so much more as the models get better, as the models get cheaper. And I think that's been probably for the startups who I'd say it's that and was like making that key decision on are
you trying to fight that battle or not? And like how much of a war chest are you going to raise if you want to try to fight that battle? And number two, it's the thing you said is embrace that this is where the world is going. You know, like we had a big thing on that. Sorry. I hear two things in that that I think are really relevant for people. And I've been hearing this advice and I like it. I didn't come up with it myself. I wish I did. But it's two things to build for. One of them is build to get acquired by a frontier company. um model ai company so do something that
they're going to need and want to buy you out that's one like really brilliant way if you could do it it's a little more speculative the other one is just make sure you're making some money now because whatever the hell you're building who knows it'll be relevant in a couple of years things are changing so fast like make sure you uh are making some money now so if you start something that does something or whatever put some money away, right? Like don't just be investing on this big equity play out because who knows what will happen.
Yeah. And, you know, look, I've seen a lot of markets. You know, I started my first company in twenty twenty and I started it literally on a business. I was trying to help people save for retirement to get their four one K match at a point where COVID happened and everyone was getting laid off. So people were like, are there going to even be four K matches if if businesses are trying to figure out how they can keep as much payroll as they can? And then, you know, I saw markets like twenty twenty one where like the money was great or even twenty twenty two where it got bad. closed up. And so you're right. You have to build an all weather business,
but also at the same point in time, it's like, you if you're building something especially in software right now and you're thinking about what does it look like over the next three six nine months if you're thinking two years you're probably looking way too far this is changing too fast right now um you know and I think you know I look at our story and it's always been just slightly ahead of where the the core market is right like we were really obsessed with gateways now the in thing is like everyone's releasing their version of a router Right. Like everyone has their routing software
and all these pieces. But, you know, I think the big thing that you're highlighting, though, which hasn't changed as a founder, you got to be fast. You got to try to be as objective as you can. And you only get so many shots on goal. So try to take them do quick, but be on to the next if you need to. And I think that has not changed. The other thing that has changed is big companies with AI. Your lead time to build something has gotten a lot smaller. You know, it's funny. I'm about to turn sixty two as we
record this. And on one level, that makes me really ancient. Right. Sixty two years young, man. It also, though, highlights some things that I think are really interesting. And one of them, like I'm hearing loud and clear, it feels like, oh, everything's changed. Let me tell you some things haven't changed. There will inevitably be massive disruptions. There will be ups and downs. And you never know if it's hitting you or not, because like.
you'd be in some part but but here's the thing that is enduring uh and I hear even now um if you get overly attached to your mousetrap idea right that's build a better mousetrap right so here's my mousetrap I'm like really loving this idea I'm going to sell it and you get fixated on your solution you will miss the opportunity what you need to do is be interested in the mice problem Yes. And the vision of a mice-free world, right, to carry the analogy further.
If you're like, you could think five years out, you could think ten, twenty, thirty years out, but you want to be thinking about this sort of fantasy ideal solution, not what the mousetrap looks like anymore, but this reality of being free of the problem, of what's possible without having this problem. Yes. fixated on the future vision for something solved, not the solution, because the solution will go through all sorts of things.
Being more interested in the problem than your solution is, I think, still a very relevant way of thinking. I think it's still the only way. I think it's like that and like, or another way to put it is the pull. People think, early stage founders think pull is this like textbook, I talk about pain, someone tells me pain and then I build on that pain. That's actually not usually how it works. it is actually someone is just like just do this thing and you might not even fully understand why they want you to do
this thing they might be like I need some solution x and you're like well you know that there's y and z that exists like no it just doesn't solve it right I need specifically this thing and it is your job as a founder to see if that pull that undescribable thing is able to be duplicated. And then if it is, you then over time get intelligent around how to explain that poll. That is not how people think that your customer can just perfectly describe this pain and like you do a user interview
and they do it. That's not how it works. And I think you're saying something that's really true, which is like, that is still the source of like, you have to, you have to look at developing products from an investigative lens. I'll give you a very quick story, but like, you know, how did we go from just building like gateway and payment software to also the agentic piece? Well, you know, it's exactly that, you know, we launched our gateway and our users-based pricing before Stripe, for example, even bought Metronome. Right. So that was like something we were doing
for startups before that even happened. And, you know, we were doing this, doing this, doing this. We did this event where we had this MCP and people were like, hey, can I actually add that to Claude? And we were like, why do you want to add this gateway to Claude? Why do you want to add our tool to Claude? Well, that's not really what it's for, right? But you don't say, don't use it like that. The thing you say is like, well, why do you want to do that? And they were like, well, Claude has connectors, but it doesn't have all of the things you guys, you guys have these ready to use tools and you guys have all these things. People were using our product to make
this agent tool better. And so that led us to say, well, then what if we just give you our product as this agent tool, right? What if we make it so that you don't have to understand Claude's system? You don't have to understand OpenAI system. You can just log into your Gmail and log into your QuickBooks and an agent already works across it. That is what our core product is. We connect to things and make it easy to use. But it's exactly that. It's not being obsessed with the mousetrap you were building. It's trying to get the bigger picture of
what people are needing. And it's not always so explicit. It wasn't just build an MCP for Claude. It was a little bit more than that. People will tell you their pain. They'll tell you things they're frustrated with. But they don't necessarily understand the nature or source of the pain. And you don't necessarily understand it. You're all just... you know, grabbing around in the dark, trying to figure something out. And definitely addressing the pain is a useful thing to do. But they also, for sure, can't tell you how to solve the pain. They just can tell you,
this is frustrating. And then there's the added level sometimes where people can't even see something. They don't even realize that they do something that's dumb. I remember coming in, I had this one really, it's so, but it's so foundational. I came in to take over running this company. some years ago. And in my initial thing, I went around and I toured the different parts of the company, the jobs. And I sat with people from a few minutes to a day or more and understand what do you do? And there was this one woman who worked
for the company and she had this job. And I thought like, oh my God, this is the dumbest job. She was fine. But the job, like why does this company ever do this? This is the stupidest, most busy company unproductive, inefficient, like, this makes me insane. In their world, they didn't think of it as pain. It just was something they did. They accepted. It took outside eyes to go, wait a minute. this is a whole job.
This is like, there's tech that solves this problem. So I said to her actually in that, in that maybe it was about a half day sit, listen, I'm virtually certain I'm going to eliminate this whole function. And I just want you to know that you have a job. You're fine. And I don't know what you're going to do next. But if you'll help me eliminate this, I'm going to give you something new to do. It'll be interesting. You'll help design it, right? But I'm going to get rid of everything you're doing here.
And I think that's the other thing too. It's funny you say it. I think there's two pieces of that that are very interesting, which is number one. The other thing is there's still so much room for people to do great work. And it's about the abundance mindset of are we caged in and saying, oh, the old work is gone. There's so much work to be done. It is only for a lack of creativity that we can't find new things for people to do. I believe that both inside of a company. I believe that in the country. I believe that at every single level.
So many things that you can still do. But I think there's a second thing you said there, too, which really is the biggest thing that I think I found on this journey, which is, you know, it started with LendTable. I've seen it with Lava. Pain is not this, like... biblically written sent from the high it's like it's like an excerpt here and an excerpt here and it's connecting the dots like no one just tells you what to build and I think a lot of times people in the early stage feel like if
I just ask enough questions I will know exactly what to build and that's never it there's a little bit more art than it than that you know it's like you're you know like a good a good chef will tell you right it's like Well, you know, the manual gets you so far. And then after that point, you know, you got to taste test a little bit. Right. You got to throw some stuff in here and there and like kind of have a little bit of understanding there. You know, I think about with lava, you know, the pain led us to would have led us to just just build connectors. Right. And it's like, you know, connectors, what we're seeing is like. every agent needs to build that sort of
thing. The real thing is, is how do you enable people, again, like you were saying earlier, the actual solve. Everyone knows that AI is making it so you should be working in a different way. Like they know they should be Working smarter. I know I should be doing it. Majority of people are still just using chat. They're just still talking to an agent. Like most of like, Bill, I know you and I are not really doing that, but that is what most people are doing. And that was the real issue, right? It wasn't how do you build the connectors? It was how do you make this approachable
where someone who doesn't understand skills and empty files and workflows and all of those things how do they end up being able to use those types of things run code execution all of those things without having to actually understand they're using those things and I think that is really the spirit of it it's it's how do you get the trinkets of pain and like all of those things and like put them together to then give something somebody something real Let me give our listeners and viewers a little thing that's useful here, because we're getting kind of geeky into
some stuff that I'm sure that some people can't fully relate to. And everybody's interested in AI. AI is the only thing with any is the only kind of headwind right now. But so our tailwind here is here's something that I'm doing again and again with my teams and that you can do now, even without my coaching. At a senior level, in your seat, and our listeners, viewers are predominantly C-suite founder
of all stages, but you're a founder, president, CEO, and their direct reports is primarily who we've got here. And in your seat... If you're not using AI about every hour of the day, maybe even personally as well as professionally, then there's work to do. So you and your whole senior team should be getting to the place where AI is of any kind, directly accessed, built into your tools, off-board, chat,
Claude, whatever, Audible, Gemini, Grok, I don't care. Like you're using it for something every hour of every day, That's maybe the right cadence at this point for every type of company I can imagine and most humans who want to actually be relevant and staying current with things. And you're not sure how to use it? Talk to AI. So for you is like, you should be using it that often. So, and if you're not,
then the conversation for your favorite AI tool is all right, this hour, this is what I'm working on. And Bill says, I should be using you. What am I not doing? What could I be doing? And AI will help you brainstorm. Give me three best ways I could do this right now. Ask me some questions if you need to one at a time. And, and you'll come up with, so if you all get to doing that, that's level one. The next level is instead of hiring a special consultant or buying some package or whatever,
just find the existing people in the company that are interested and give them like a weekly or monthly, weekly is probably best at this point, a weekly like lunch and learn kind of thing. mastermind, whatever you want to call it. And they get together and they share with each other something cool I did. So the first round is something cool I did with AI this week. And that will cross-pollinate interesting ideas and breakthroughs. Mitchell will share with me something. I'll share something. I'll be like, oh, I like what he did. That'll spur me on to do something similar
or related in my world and vice versa. And the second one is... Look, something I'm struggling with and everybody shares something they're struggling with and then people will naturally see some ways to connect tools and things like that. You will naturally start to get more and more people until it's just an obvious part and it's fully integrated in every part of your company. And but just go with the people who are already enthusiastic. Right. Just lean in. Just buy them lunch. You know, simple as that. Get them together. Buy them lunch. Let them have their little sort of
informal hackathon. And that's like you're leveling up, you know. And we call those the flag bearers, right? Like there's a flag bearer in every company who's gonna like push the company forward. And those people now more than ever are super important, right? Like now more than ever discovery, just playing with new tools is very, very important, right? Like that's one of the things that people find about Lava. Like, you know, one of our customers, for example, What she does, she runs a retail clothing company.
She took her HubSpot, combined it with her Stripe logs to then do an outbound like physical mail campaign with another tool called Lob and did all of that through Lava. And so she's not an engineer. She's not like a product manager. She's very, very smart. But what she understands is she just talks to Lava, asks, does Lava have this tool, this tool, and this tool? It'll say back, yes, I do. And then she asks, can Lava then just use this tool to
do this and this tool to do this? And it's like, yes, would you like me to do that every single day, every single week, one time? And that's truly where we're at. And the most important thing, and Bill, I think you highlighted this well, Do not be afraid to just ask a question. I need to get X done. How could I do it with this tool? If the tool I'm currently using cannot do it, what tools can? That is where we're at now. And it is super important for folks to understand it is that easy.
And it might seem scary. It is truly not. And you asking questions is probably eighty percent of the battle. I'd say it's probably about eighty percent. What would you say? Yeah, I think it doesn't really matter where you are. Just get started and just start playing with things and just do more and more. And that naturally leads to more and more. And then if you find yourself a little in a corner, then you want to ask somebody or get some help and dig out or try a different approach. I had a client recently who
How long has it been now? Six months, maybe? And they were having all kinds of software issues. And they were having AI adoption issues because they were running into problems with the particular AI tool they were using. And it wasn't integrating well with the company's other tools. And I'm like, oh... I do know the problem there. That company is way too big and they don't know how to talk to each other and their tools don't work together well. And it might someday be the best because they own more data than anybody else. But it's not currently.
And the problems you're having are very, very... So try this other tool. Now I'm going to leave both of those tools out because I don't want it to be about that. But... And they're like, okay, I'll try it. So we did a little bit of work together in our, in our scaling session. And then I went away and like a month later, they're like, oh my God, like roadblocks are being removed. This other one works so much better. I'm like, yeah, you need a little outside perspective. You need to be agile, flexible in your approach. Cause sometimes you find yourself boxed into a corner and you're like, oh,
we committed to this one because of that. Yeah. On the surface, that's good. But if it doesn't integrate well, and tools like yours that are letting people do more across more things are helping them not to get boxed into a corner, right? So whether it's talking to a friend or trying some different tools, that can really help. Because there's times I had a thing I was building with Claude and my results were getting worse and worse and worse. And I was like, what the hell and then pretty quickly a few hours
like spooled up and I was completely unproductive like all morning and eventually I'm like I'm stopping. This is making me insane. And I want to punch the machine. And it does no good, right? So I stopped. I went out. I went surfing. And I was out on the bay. I do this thing called wing surfing. And I'm out on the bay on the Golden Gate Bridge. And a couple of things happen. It's really beautiful. And then I suddenly realized, oh, the way I'd organized the job was wrong. and uh claude wasn't helping me reorganize
it and I came back and I changed a little bit the structure and everything was fine right so sometimes you need a different perspective you need to step back you need to try something else try a different tool like whatever you know so giving yourself some room trying some things don't getting too stuck you know and don't become too convinced it isn't possible because it maybe is just a shift in your approach Shifting the approach is normally the right thing. And the other thing is every agent is connected to the internet. Tell it to go do some research and
usually it'll also come back. One of the things that people often don't do is don't just take its first answer, especially if that answer came to you in under like two seconds. Tell it to go back and do some research. Just for folks who maybe are newer to agents, there's two modes in which agents usually operate in, a fast mode and a slow mode is the way you can think about it. The slow mode is called the thinking mode. and they're naturally trying to prioritize getting you something back quick um oftentimes though what's the expense of getting you something quick the same thing
as if you ask you know one of your direct reports to get you back something quick right they're gonna take heuristics they're gonna take shortcuts they're gonna say well I mean I checked this like a month ago and it was this could have changed in a month right like those are all the types of things that like just you Tell it to go get a second opinion. And oftentimes you will find that there's actually a way to do the thing that you're trying to do. But I think the most important thing and I think it seems like we're aligned on that is like first step is just try some stuff, right? Like just just talk to it. And, you know, I think, you know,
companies like ours, our goal is to make it so that you can talk to an agent and get more done than you ever thought you could. And, you know, I think that there's levels of complexity. Some have a lot of bells and whistles. Some try to strip it down. Our goal is to make it so that, you know, every time you're doing something, it's just in the tools you already know and you already understand. Right. And like that is usually the thing that makes it pretty approachable. But there's different levels and different flavors. The most important thing, though, is find something you're really, really good at. Find something that you trust. And then finally,
it's like if you like how it feels and it's getting the job done for you, if it ain't broke, don't fix it. I've always been a big believer in that as well. Well, let me highlight another recent little mini example of this. So I wrote this book, Busy is Broken, right? Go read it. Go buy it. You're already sending it to me, right? Probably. And it pointed to a gap in the research around CEO behaviors and what's good and bad and what gets in the way and the right habits for CEOs in
particular. It actually applies to other leaders, but it's really, anyway, so I decided to, I was beginning a PhD program. And so I decided to do a study on CEO behavior, the hours worked and how it worked, including talking to the CEO, but also talking to their direct reports, the people that report to you. So I designed the study and to be like academic and rigorous and all that, I built it over in this platform called Qualtrics. And Qualtrics is like a research and
survey thing. It's very complex. It's old school. And it's definitely kind of a pain in the ass to program if you're not doing it all the time. And I'm not like first time using Qualtrics. And so I, so I have Claude talking me through it and I'm just like, oh, like it's describing things and they change and the menu doesn't look like that. And I'm trying to like, and it can't script it. And it like, oh, I was just losing my mind. I got the basics of it working. By the way, you can go check it out if you
want to try it and so on. You can participate in it and you can contribute to the research, but it'll also tell you interesting things about your leadership. You can go to scalingcoach.com slash study, scalingcoach.com slash study. And you can be a part of this PhD research into CEO behaviors. And it'll tell you some interesting things about yourself and how you rank to other CEOs. But you do have to tell us and give us some emails of a few of the direct reports so they can give you some anonymous feedback. So that's the study. So I was having all that trouble with
it, right? So then I said to Claude on a revision, I'm like, okay, I'd like you to look at the study. I'd like you to tell me where it doesn't flow, where it's confusing people and help me clean it up. Oh, you know, since you're already there, maybe you can do some of the stuff. So I had it evaluate and critique the thing. And then I had it go fix it all in Chrome, Claude driving Chrome. oh my God, the difference in my spending days getting something done versus it being done in two
hours, it just fixed, oh yeah, this is that, the error checking in the page and the thing like that. so much improved like user experience and better participation everything like that and it just did it like and I'm like why did I do it the other way like why was I hand coding that it's it's it's funny man it's like you can truly I think that this is one of the most exciting times to be really just doing any job right in like the sense of like
Yeah. Any like anything you're doing, you know, if you're if you're sitting at a computer, you probably can make whatever you're doing a little bit more efficient. And, you know, I think for us, our goal is always to say, like, you know, The most important thing we can do for people is make it where your day is just a little bit more delightful. Like that annoying task that you always had to do by hand. Like, how do you now start focusing on something that helps you climb the ladder? Because you can do some thought leadership. Like, how do you start to go to those things? And I think that is the thing that
with when you look at like the industry, I'm very, very excited about that piece. It's just like. I think that in every era when humans have had more time on their hands, cool things have come from it. There's been a lot to figure out in every single one of those eras, but I think we're vastly getting to the point where ideally we're going to have a lot more time on our hands and that's usually when some of the coolest, newest, freshest things end up getting built and usually they push what we're doing as humans forward. I think that is completely right.
It's there's just so many ways to do some of the stuff that like, you know, like Qualtrics or these old legacy tools, right? Like that's actually what our specialty is. Like you tell us an old legacy tool you use. We know you're going to, yeah. You just, you tell us a tool. It probably already is in our connector. If it's not, We can get you up and running with it in literally hours. So I think that's the North Star, is how do we enable every single business, which they're not going to rip and replace every single old tool. That's silly. Those tools are enterprise ready.
They're secure. They're safe. If anything, what has to happen is we You know, there's startups like us that don't think about it as ripping the old tools out. We think more about how do we fit in with the existing tool sets. And that's really where we think that the future is going to go. I don't think that Salesforce is going away anytime soon. I don't think Qualtrics is going to go away anytime soon. I don't think SurveyMonkey is going to go away. Like what those businesses are going to do is new businesses are going to figure out how do we tap into the infrastructure
you've built that's secure, it's scalable, all of those things. And how do we push that forward? That's really what we're expecting. His company is Lava.so. Mitchell Jones, thank you for coming on the show and dropping your stories, knowledge, and experience on us. Thank you so much. Thank you. Appreciate it, Bill. Thank you. Everybody go check out Lava.so. We help you get your best work done directly in the tools you already use with the Lava.so agent. Thank you. And as this show drops, that study is closing. So if you're curious about the PhD study,
scalingcoach.com slash study, go participate in it now. You'll get really interesting feedback on your leadership style and ranking relative to other CEOs in the world. If you want to know the book, busyisbroken.com or just go to find busy is broken on um what amazon wherever you get your books in every form you could imagine kindle audio etc um thanks again everyone for watching for listening um it's been a pleasure being with you um thanks to my friend and mentor vern harnish who created the whole scaling up framework that's the heart of
our work um and to Claude for helping us get our show produced and distributed and prepared out every week thanks again for watching for listening until next time keep scaling
In 2006 Todd Palmer was $600,000 in debt and about 60 days from closing his staffing company. He turned it around and made the Inc. 5000 six times in seven years. It never felt like winning. "All it did was leave me...
This year Bill will take about 80 flights, launch a book, push his PhD research into phase two, coach clients on three continents, and still finish with close to 90 days of wing surfing. So why doesn't his busy feel...
On January 4, 2009, Bill Flynn's first official day as general manager of an email hosting company, the entire technical infrastructure collapsed. For two and a half days, customers couldn't reliably send or receive...
American Airlines flies about 1,800 more flights a day than United. Last year it made about $3 billion less. More motion, less money. Bill was an Executive Platinum flyer on American until the delays, broken connections...
When Apple announced it was opening retail stores in 2001, analysts gave the idea two years before the lights went out. Twenty-five years later, Apple Stores are the most profitable retail spaces on earth - more sales...