Subscribe
Share
Share
Embed
AI First with Adam and Andy: Inspiring Business Leaders to Make AI First Moves is a dynamic podcast focused on the unprecedented potential of AI and how business leaders can harness it to transform their companies. Each episode dives into real-world examples of AI deployments, the "holy shit" moments where AI changes everything, and the steps leaders need to take to stay ahead. It’s bold, actionable, and emphasizes the exponential acceleration of AI, inspiring CEOs to make AI-first moves before they fall behind.
Andy Sack (00:01.695)
This is AI First with Adam and Andy, the show that takes you straight to the front lines of AI innovation in business. I'm Andy Sack and alongside my co-host, Adam Brotman. Each episode, we bring you candid conversations with business leaders, transforming their businesses with AI. No fluff, just real talk, actionable use cases and insights for you.
Welcome, everyone. Super happy to have Benji from Galley Solutions on as a guest. And let's dive right into it. Benji, do you want to give a brief introduction of yourself and your company?
Benji Koltai (00:44.098)
Great. Yeah. First of all, thank you, Andy and Adam for having me. So my name is Benji Kolthai. I am the co-founder and CEO of Galley Solutions. My background is as a software engineer. Went to school, got a computer science degree, started working in tech in Silicon Valley, and was also diagnosed with Crohn's disease in college, an irritable bowel disease, and essentially was told, doesn't matter what you eat.
You're going to have seven to 10 surgeries throughout your life. You just kind of have to deal with it. We don't know how it works. So I continue to live my life after school. When I went to start working in tech, had my first surgery, eight inches of my intestine removed. And after that surgery, that was kind of my wake up call to say, hey, you need to start maybe making some lifestyle changes. I actually met my now wife who was eating gluten free and she said, hey, you should probably worry about and care about what you eat.
try gluten-free, it'll help. So I started eating gluten-free and that was really the start of the galley journey. I ended up joining a food tech startup called Sprig, which was one of the delivery only restaurants in San Francisco. And that was a full stack food service company, the Uber for Food. And it was there that I discovered the problem that is an analog kitchen. And so Sprig had a large commissary kitchen. I, as a consumer, was trying to use the app, which would label things as gluten-free or not.
And frequently that information was wrong. And when I asked why it was wrong, I found out that it was humans checking check boxes. But gluten-free is something that you can calculate from a recipe. So I went to ask the culinary team, where's your recipe management system? They kind of chuckled, opened up a Google spreadsheet. And I was sort of shocked that this large commissary kitchen that was super sophisticated and high tech was using a spreadsheet. And that's really when I discovered that the food service industry is run on pencil and paper. It's run on spreadsheets.
kitchens, for the most part, really don't have a system of record, a source of truth for their core food data that consumers like me who depend on that data for their health can rely on. And so I started building the first version of Galley at Sprig. And when Sprig shut down in 2017, I spun out that solution as a standalone solution. And we are a recipe first culinary resource planning platform.
Benji Koltai (03:09.058)
which essentially means that we're trying to organize the world's food data. So our vision is to improve people's lives by optimizing kitchen operations. And we have a strong belief that by digitizing kitchens and recipes in the food system, we can improve everybody's lives.
Andy Sack (03:27.707)
Awesome. Thank you for sharing such personal information as a founding story of a company. makes a lot of sense. I'm curious, Benji, what's the status of the company today?
Benji Koltai (03:45.506)
Yeah, so we raised our Series A in 2022. So, you know, we're launched, we're live, we have close to 200 businesses using our platform. Broad range of customers from single unit mom and pop restaurants to large contract food service, commissary kitchen, delivery food companies, pretty much any kind of food service company uses our product.
Andy Sack (04:14.069)
and the product does what?
Benji Koltai (04:16.802)
Yeah, so it primarily works to make the process of running a kitchen more efficient. So we start with recipes and then we have a planning module. So you can say, hey, this is how much of each thing I need to make next week or for the next six weeks or tomorrow. And then the process that follows from that is inventory, procurement and production. It's the same process anybody who makes food does. You as a home cook do it.
You think about what you want to make and how much of it you want to make. Those are the inputs. And then you have to decide what do need to buy? What do I have on hand that I should use? And then what are the order of operations that I need to do to actually make that food? And so that's really what the platform does is it has those workflows and those modules to allow anybody to make food at any scale.
Andy Sack (05:03.477)
And what's the role of AI in all of this?
Benji Koltai (05:07.714)
Yeah. So we started out with sort of an AI focused vision. The whole idea is that AI is coming. AI is here now. AI was coming when we started. And the premise is that the kitchen is sort of a incredible juxtaposition of industrial engineering and art. And it's such a craft to be a chef to
make delicious food, but also have it be consistent and have it be efficient and have it be profitable ultimately. And so the whole idea is that if we can digitize that process, we can accomplish all those things. But in order to digitize it, especially for this type of user who is not in front of the computer most of their day, or at least they don't want to be, you need to leverage the latest technology that improves the human computer interface, which we have always believed would be AI.
We want to have the human do the least amount of working on the computer, interfacing with the computer, and have the computer have a rich enough schema and understanding of the domain, but then also be able to apply intelligence, both as an interface for the human, as well as a decision engine underlying that whole process to help the human make the right decision in every moment. Because in food, every instant, your next step can change. It's kind of like chess.
with a thousand opponents rather than just one, because the heat might be off or the ingredient might be different. And so you have to make very fast optimization decisions, which we from the get-go knew would be a perfect use case for AI.
Andy Sack (06:51.669)
Adam, let's get you in here. What are your questions for Benji?
Adam Brotman (06:56.552)
Yeah, first of all, Benji, it's fun to have you on. I remember Spreg, I actually visited your, I don't know if you had multiple, but you had, I remember like in, I want to say in San Francisco, you had like a commissary. And when I was at Starbucks, we did like a field trip and came and visited you guys. We were just sort of fascinated ourselves about the whole, it's like when DoorDash was kind of coming up and you guys were doing your thing. I, Spreg was super cool.
I guess my question for you is like going back to the AI part of it. you now like, give us a little bit of of your, of a sense of your journey. Like you, you've been building this since 2017, right? So like you're, you start building it and you're using my guess is traditional full stack, um, uh, software, uh, and databases. And you're like, Oh, like let's digitize this and let's database that and let's come up with.
software that's somewhat deterministic to help with this journey. then at some point you realize that there's a bunch of researchers that are building a non-deterministic, you know, general intelligence neural network system that actually is starting to work. My guess is you couldn't have discovered it much before 2017. But and probably if you're like the rest of the world, you probably really started to pay attention in 2022 and
when everyone else did at the chat chibi team moment, like, give us a little bit of a sense of like, when did you have your aha moment around generative AI, non-deterministic AI, and how you could apply it to what you were already building at your company?
Benji Koltai (08:41.934)
Sure, yeah. So in 2017 when we were starting, flavor of AI, like if the flavor of AI today is LLMs and generative AI, the flavor in 2017 was big data and data schemas and things like that. And so that was the flavor under which we sort of nestled in the AI concept. So my seed investor is an AI first investor called Zeta Ventures.
and they were very early in the AI push. And my big pitch to them was more that data schema. And so in the beginning, it's a CRUD app. It's, hey, how do you get all the inputs and outputs and build this very standard sort of CRUDs, know, create, read, update, delete SAS product and get the schema right, get the domain language, get all of the pieces of the puzzle right. Knowing that having a well-schemaed
API first is another key piece of the puck that we were skating towards work. Every piece of our platform is available in our public GraphQL API, which has now been a really great choice for plugging into LLMs. So to answer your question, the beginning was AI to us meant a really strong data schema, capturing a lot of data.
and being very sort of API and external interface friendly. And then ChatGPT happened, it was like, okay, great. Here's the next flavor of AI that we are. And we put out a blog post where I was explaining that we are sort of AI-ifying the food domain. And so the big innovation in LLMs is predict the next character, predict the next word. In video and image, it's predict the next pixel.
In our domain, it's predict the next action. And so how can you create a platform that has enough data, sort of human generated data that says, hey, when all of these things happen, the next most likely thing to happen is this, so that we can start to build the data set that an AI can predict the next action on behalf of our operators. And so I would say that's like the high level concept. And then we've been implementing LLMs as product features. So
Benji Koltai (11:06.446)
our data inputs and data enrichment workflows are leveraging LLMs currently.
Adam Brotman (11:15.465)
So just so I understand your product a little bit better, and for our listeners, most of your customers are like, are they CPG companies that are making like food products that are gonna get sold online or in grocery stores and whatnot? Or are they mostly like restaurant chefs and on behalf of a chain or even a big individual restaurant that are like trying to
Kind of explain to us like how your customer uses your product, like just at a very tangible level.
Benji Koltai (11:49.762)
Yeah, sure. So first of all, think, you know, college and university dining halls, prepared meal companies that do, you know, delivery meals, hospitals, B &I, you know, corporate campuses, things like that. So that's our core ICP catering. Restaurants are included in that. And then in terms of the workflows, you know, a lot of those types of businesses that I just described have
Adam Brotman (12:10.974)
Got it.
Benji Koltai (12:18.488)
high menu variability. They have ahead of time knowledge of what is going to be served. So like think a caterer, right? There's, you know, 20 events for a month and every event has a different menu. It has a different number of guests. It has a different time. And so getting all of that coordinated in a kitchen and having one kitchen, maybe produce 20 events worth of food in a given week is a very complex task. And so understanding
Adam Brotman (12:45.769)
Yeah
Benji Koltai (12:46.54)
What do I need to buy? What do I need to have on hand? What do I need to make? Maybe I make all my chopped onions once and I use a quarter of it here and half of it here and a quarter of it there. So getting all of that planned out.
Adam Brotman (12:49.158)
I see.
Adam Brotman (12:57.961)
Well, that's kind of cool. like, other words, I tell me if this is right. Your core customer is like, I'll call it like this industrialized kitchen that serves like a whole hospital or a whole school or whole campus, whatever, and or a caterer. this is like a ingredient intelligence system that's going to be like, hey, I'm going to reverse engineer. Like, you're going to think about the menu and the items that ultimately are going to get served to X number of people on X date.
And I'm gonna take all of that and your system allows them to like bring it all the way back a month ahead and say, to be as smart about buying and preparing and storing your ingredients for the next 30 days, I'm gonna take the end in mind and I'm gonna like use software to tell you what to buy now so you don't have to like plan it out. Okay.
Benji Koltai (13:48.064)
Exactly. And, hey, know, John just burned a batch of sauce. Here's the quick rejiggering and remediation of how we're going to pull forward your inventory. We're going to place this purchase order. We're going to help you recover from that inevitable error.
Adam Brotman (13:56.029)
Right.
Adam Brotman (14:02.739)
So can I, so sorry, Andy, one last question. So I'm going to ask you a provocative question, a fun question, which is like, you know, is it, did, was there a moment where you're like, you've been working on building, using big data, big data and the schemas putting aside the API for a moment. And you were like building all this stuff, like the hard way. And then did one day, did you just go, shit, like
Andy Sack (14:04.723)
Yeah, no, it's helpful.
Benji Koltai (14:11.416)
Okay.
Adam Brotman (14:30.459)
I could actually just give this to like a really sophisticated LLM thinking model and it can, which is great. And they can like just do this in a way. And maybe all the stuff we did, we could feed into it. Like, was there a moment where you're like, let's not fight it. Let's just embrace it.
Benji Koltai (14:44.62)
I still remember it. was earlier this year. I think it was like January 4th, maybe of this year. Where it's like, all right, the LLMs are here. The models are sophisticated enough. Let's just try what happens. And honestly, one of our biggest challenges is getting the recipes out of people's heads and into the system. And so that was our first use case. And I just threw it at the LLM model, which at that point had just come out with their JSON schema support. So essentially now we can use this agent or this LLM.
Adam Brotman (14:50.355)
Great.
Benji Koltai (15:13.88)
to take the infinite set of recipe data, whether it's a handwritten grandma's recipe from your recipe book, or a video of a recipe or a URL from a blog post and say, here's your target schema, go ahead and turn it into that schema. And it was like, wow, this works, this is amazing. And so we can demo it on sales calls now where you can just type in a prompt and it will use your catalog that is your unique set of recipes and ingredients and generate a recipe that aligns with what you're asking it to do.
Adam Brotman (15:43.275)
I could keep going, Andy, do you me to ask? So along that same line, Benji, because that's fascinating. A lot of our listeners are CEOs and leaders and they're running companies kind of like yourself. most of them are like consumer brands, you know, but we have a good mixture of B2B companies and software company CEOs. And there's this one question that comes up all the time, which is like, am I...
Andy Sack (15:44.479)
Yeah, keep going.
Adam Brotman (16:11.91)
How am I going to, am I competing against my competitors? Am I competing against the LLMs? Like if you're a software provider at your core and these LLMs can just instantly and more and more the case produce any software. there a moment where Benji, it's a two-part question where you were like, uh-oh, like how do I maintain a moat or a competitive advantage when anybody can just do this in an instant and kind of explain your thoughts on like, you know,
Benji Koltai (16:16.141)
Mm-hmm.
Adam Brotman (16:40.936)
I'll call it like the other things besides the software that are really important to maintain a competitive differentiation.
Benji Koltai (16:47.18)
Yeah. So definitely have been thinking about this almost nonstop since ChatGPT came on knows, know, SaaS is dead and all these things. And I think that there are certain types of SaaS that will be replaced entirely by these vibe coded, you know, AI driven apps. But what AI doesn't have yet at least, and honestly, given the current architecture, I don't know that it will have it, is the deep
sort of statefulness and the rich, nuanced, domain specific context and domain language that is that sort of language that I talked about, you know, sort of equating predicting the next pixel to predicting the next action. And so for a food service company, having an LLM be your culinary resource, you know, essentially think we, we called it very explicitly CRP.
analogous to ERPs, where an ERP represents the moment by moment state of a business. Usually it's financial. There's MRPs for manufacturing resource planning, where you have all your bills of material, all the widgets that you make. And so the core concept is that a system of record, in my view, is the most defensible SaaS in the world of AI apps, because AI does not
have that level of storage. It can leverage that level of storage, but it still needs to have a central system of record that maintains all that information. And by the way, that system of record needs to be usable not just by AI, but also by humans and also by other systems. And so we've been building this platform that is this data platform that currently has a web app for humans to interact with it. We are building out that
MCPs in the chat interfaces for AI to interact with it. And our belief is that the whole ecosystem is going to move into these AI to AI interfaces. And we need to maintain the determinism and the statefulness that is the kitchen so that everything doesn't just vanish when the LLM's context gets corrupted.
Adam Brotman (19:03.22)
Yeah, that makes a lot of sense.
Andy Sack (19:03.887)
That makes sense. Let me ask, given the caterers and the hospitals that are your customers, what's the value prop that you sell to them?
Benji Koltai (19:17.74)
Yeah, so reducing the chaos, which ultimately leads to saving money and increasing revenue.
Andy Sack (19:27.327)
So in its essence, is it an inventory management system?
Benji Koltai (19:33.606)
no, I would say in its essence, it is a recipe management system, first and foremost, and inventory can flow from that, but so does nutrition analysis. So does costing. So does planning. So there are a lot of inventory management systems out there in the food service industry, but in our view, that's like a third step to the problem. And if you want to be plugging in your sales data,
Andy Sack (19:56.447)
Yep.
Benji Koltai (19:59.008)
into that inventory management system, then you better have a strong representation of what selling one burger does to your inventory, because you need to know exactly how many pounds of onions you're consuming when you sell a burger. And so the ROI is reduced food waste, reduced food costs, improved labor efficiency, as well as operational scale. We have had multiple customers
who started with one commissary kitchen and that was overwhelming in itself to run. And then when they migrate on to Galley, turning on four more commissary kitchens is a couple clicks of a button. And all of the operational things that need to get set up to run that kitchen are handled by our system.
Andy Sack (20:46.165)
How big is your company? How many employees?
Benji Koltai (20:49.604)
We have about 30 people right now.
Andy Sack (20:51.829)
And how is your company using AI?
Benji Koltai (20:55.17)
Yeah. So, I have been pushing everybody to always ask, how can I accomplish this task with AI first? And you might try it six months ago and fail and like, that did a terrible job. And there are two things there. All right. One, maybe the LLM isn't smart enough yet, or it doesn't have enough context, or we haven't wired things up enough for it to actually do the thing. And two, you as an employee, haven't learned this brand new paradigm and this brand new tool.
that is going to take practice and it takes a lot of failure in order to be successful at. And so just constantly pushing my team anytime a problem comes up, anytime there's cross team communication of, know, one thing that I've been loving is our CS team used to ask our product engineering team, hey, how does this logic work? you know, when a user goes to this page, what's the permission that drives whether they have access to this control or not?
And so I've given all of my CS team access to our GitHub repo and a cursor account. And I've said, go ask cursor first before you ask the product team. And the amount of sort of friction reduction in empowering each individual to find the answers, given the digital knowledge that we've accumulated over the last eight years, has just been incredible for efficiencies and improvements.
Andy Sack (22:15.733)
Do you use predominantly?
if so which one?
Benji Koltai (22:22.594)
I personally have been using cloud a lot. I still write a lot of code, so, I use cloud code. but everybody I've sort of told people, you know, don't let LLM licenses inhibit your experimentation. Try new things anytime, you know, chat, GPT came, GPT five came out. I tried that for a few days. So, this is constantly changing and I'm constantly experimenting.
Andy Sack (22:48.403)
And as the organization, is there one that gets used more than others or it's across the board?
Benji Koltai (22:55.0)
I think it's probably across the board. have some shared GPTs and chat GPT that are useful. But for the most part, I think everybody kind of experiments and finds their own workflow that works best.
Adam Brotman (23:08.234)
How do you keep up the encouragement? I mean, I love what you just said. Like, hey, ask AI first, know, try using AI to either solve your problem or get after something before you escalate to the rest of the organization to help you solve the problem. Do you find as a CEO that you have to, do you formalize that process in any way? Do you have like a AI task force? Do you do office hours? Do you do?
Contest, mean, what kind of mechanisms do you do to reinforce that message you said you'd give to the company?
Benji Koltai (23:42.146)
Yeah. So first of all, I try to model it. I mean, to me, that's my number one leadership style is if I'm asking people to do it, I better be demonstrating that I'm doing it all the time. And so I'll post videos of, hey, here's a workflow that I was able to accomplish really well using AI. We might have a hackathon. That was a big thing to get my developers on board. I kept harping on the developers. You need to be using coding tools. You need to be using AI tools.
At our most recent offsite, we did a hackathon and I essentially made it mandatory at the beginning saying, hey, I'm going to be walking around the room and I want to see how you're using your AI tools. And a lot of people who weren't using AI coding as much were kind of put on the spot. And at the end of the hackathon, lot of the retroing that we did is, wow, I totally underestimated how big of an impact using an AI tool like this would be.
Adam Brotman (24:40.861)
Interesting. I love that.
Andy Sack (24:44.713)
Have your customers, like talk about how your customers sort of either understand or don't understand the solution that you're providing to them as it relates to AI.
Benji Koltai (25:00.866)
Yeah, so it's funny, like a year ago, we've had customers who've said the AI in Galley is amazing. And what they were referring to were the math equations that we were using to calculate how much they should be purchasing, where we do the summing of their demand and we reduce what they have on hand. so AI is a very broad concept that I think most people just kind of translate to
computers doing smart things. And the range of things that our customers perceive as AI is, in my view, appropriately broad, because I would agree a math equation implemented as an algorithm is technically AI. It's artificial intelligence helping you understand what to buy. Now, as ChatGPD continues to become more consumer and people are used to being able to chat with a blog post or chat with all various things, more and more our customers are asking, so,
Can I ask a chat bot in Galley, hey, what's my food cost for next week? Or what are the biggest drivers that are driving up that food cost? And so we are starting to get more and more people getting interested in having those chat-based interfaces to their data, as well as having the expectation of what we've implemented, which is I don't want to have to click all these buttons, fill in all these data points, like...
I want to just be able to take a picture of my recipe and have it upload to Galley and have it work and not have to be this special schema. And so there's a lot of sort of aha moment or alignment with our customers who everybody's asking, how can I implement AI at work? And we provide them with a very tangible example of what that means and what that looks like.
Andy Sack (26:47.061)
Prepping for this episode just prior to recording, we talked about some of the philosophical elements of AI, and it was an area that you wanted to touch on. Tell us more about that, or tell the audience about that.
Benji Koltai (26:58.604)
Yeah.
So yeah, sure. So honestly, like for me, one of the biggest first hurdles was how can I align my why, my personal why with adopting AI? How does the statement, I am going to adopt AI align with my personal why? And it comes down to adaptability and staying sort of on top of the latest technologies.
And grappling with the trade-offs, everything has a trade-off. Every solution is trading problems. And so what problem is this trading in me adopting AI and using AI more? Well, you know, there are all the potential negatives of it's going to take my job, it's going to make me irrelevant, it's going to make my company irrelevant. And for me, addressing those head on and thinking about them critically for hours helps me
find the path forward and find the middle ground that says, well, yes, it probably will kill a lot of SaaS companies. And yes, it does represent sort of an existential threat to my career path. And so how can I become OK with that? How can I find harmony and figure out how I can productively bring AI into my life and into my company and orient my company towards this thing that is inevitable?
me choosing not to adopt AI is not going to prevent AI from taking over the world. And so I've had that conversation with myself and then I've also found it really important to have that conversation with my employees and say, know, AI is scary. AI might take your job. So what does that mean for your action today? What does that mean for how you want to be in three years?
Benji Koltai (28:54.52)
Do you want to be a Luddite who has just said, I'm not going to touch it because I think it's the devil and it's going to take my job eventually? Or do you want to become the cutting edge employee that everybody wants to hire because you have mastered this new tool that requires human interface and control and collaboration? And I think that it's really digging down into those whys. And as an executive trying to bring AI into my workplace, encourage my employees to use AI.
I think that if you try to do that just of like, you must use these AI tools, I'm going to force it on you. And you don't really talk about that existential, philosophical, personal, emotional reaction to what it is to bring AI into the workplace. I don't think you're going to be successful because ultimately we're emotional beings and we need to be having those hard conversations.
Andy Sack (29:49.237)
Wow, Benji, way to tee it up. I mean, I'll say that it's really, I mean, I don't know if it's the first time, but it feels like it's the first time that we've actually discussed that you've teed up really the existential philosophical dilemma that AI poses to all of us. Like Adam and I periodically.
ingest, be like, oh, yeah, AI is smarter than me, and it's coming from my job. But your articulation is really well stated. I find myself in quiet moments thinking, really, the existence of Form 3, is really about digital transformation for the AI era, helping
business leaders, particularly CEOs, transform their businesses to adopt AI. And underlying that is a tremendous amount of uncertainty, unease. Is it the right ethical questions? And so I really appreciate you bringing it up. Adam, you have any reaction to what Benji said?
Adam Brotman (31:03.818)
Yeah. First of all, Benji, I thought that was a really great articulation from a human perspective, from a CEO perspective. I actually feel very similar to you. I actually, don't, Andy and I similarly, we talk about this topic when we talk to business leaders. And I think you touched on something that is very true, which is like, you almost have to compartmentalize for a moment.
your own feelings of anxiety and fear and unease about all the unanswered questions and ethical questions that come up around this technology. And this technology is happening. I like the way you put it. You can't...
you can't possibly put it back in the bottle. so it's a tool that is out there, is gonna be infused into everything. And so you can't just stick your head in the sand about it. I do wish there were more conversations like this, like you just teed up that you said you have with your employees and with your friends and family. And I think we need to be doing a lot more of that than we are right now. I think right now, to be honest, it's moving so fast that
It's like hard to get your brain around it, right? Like you, I was just actually talking to a group of parents at a school about AI and this was exactly the topic we were talking about, which is it's almost like we're still in a bit of shock even for Andy and I and for leaders like you that have been in software. it's, it's shocking what it can do. It's shocking how fast it's happening. And there's like multiple different, I'll call it companies and countries that are racing now. So,
All that being said, I agree that like the best thing we could do is actually understand it and talk about it and not shy away from that. being, I call it being anti AI just, just for its own sake doesn't help. But I do think having honest conversations about what's scary, what's unethical, what's
Adam Brotman (33:11.468)
And what we might be able to do about it are important conversations to have. I just think that maybe we could all, that is something we can do about it. So I really like what you said, Benji, a lot.
Benji Koltai (33:21.742)
Awesome.
Andy Sack (33:23.497)
Benji, anything else you want to add on that topic? mean, it's a deep topic. Anything else you want to add on it?
Benji Koltai (33:31.502)
I don't think so. mean, I think, yeah.
Andy Sack (33:34.687)
Yeah, no, I think you framed up well. Adam, anything else for you? And if not, any last questions for Benji?
Adam Brotman (33:45.712)
Um, no, no, last questions. I thought I it's interesting. Yeah.
Andy Sack (33:49.875)
Yeah, so let's transition and Adam, why don't you summarize the highlights. What do you want the audience to take away? What are you going to take away from this conversation with Benji?
Adam Brotman (34:01.93)
Well, what I love about this conversation and it's unusual for us to have, we usually work with B2C consumer brands. and we, and we talk about like what the application layer looks like, at the, at the, at the end of the, of the edge of the network, like, you know, and as we go closer in towards the B2B companies, the software providers, and even towards the models.
we don't usually have those conversations. And what I love about this conversation is that you're sort of a really neat mix of somebody who is mission driven, is very thoughtful about helping solve a business problem by
using what we used to, know, kind of old fashioned AI. Now you've got this sort of blend of like data and old fashioned AI with LLMs. but what I, what I, what I wrote some notes on were, I love how you, you've been thoughtful about, okay, how do I adopt generative AI in, in, neural network AI into the solving the same problems that you were trying to solve even before?
this became on the scene. So, and, you've been thoughtful about, like, you mentioned something I thought was really important, which is that the, the data itself that gets compiled and understood by your platform on behalf of your, and your customers is something that, the, LLMs in, in, or the data, these AI systems in general, they're not going to have access to that specific data.
And your ability to sort of like adapt and realize that and say, okay, great. This is like, instead of being afraid of what generative AI could do to a company like yours, you were just like, fine, it's going to be augmentative, it's going to be enhancing. And it doesn't change the problems we're trying to solve. And it doesn't change the importance of the unique data you have. I think it's an eye for me, that's an important lesson.
Adam Brotman (36:17.707)
Whether you're a B2B software company or a platform or you're an end user consumer brand, feel like Benji, your process of, of adapting to this new technology and using it, like almost jujitsu-ing it to like, yeah, I'm going to use this in my favor. I'm not going to take my eye off the ball of the data that's important to me. I'm not going to take my eye off the ball of the problems I'm trying to solve. I feel like that's a leadership lesson that
is one that our, our listeners could, could really build off of. I thought you gave a specific example. I'll just give this other example and then turn it back over to you, Andy, which is you gave a specific example that I love and I want to highlight, which is that you said. We've got, I think it's because of your computer science background that you mentioned. So like you were like, Hey, I'm in the business of providing this, you know, kind of complicated schematic software that ultimately makes my.
that makes my customer's life easy. You're in the job of taking a very complicated set of ingredients and processes and making them easy. And so I'm imagining your customer support representatives that you mentioned, they're probably getting questions like, how does your system work in this way or that way? And you said, I could just let, I could use the power of an LLM to just let my customer service representatives just talk to,
with Claude to basically talk to the system, talk to the software almost. Like here's a software system that knows how to instantly peer into the software and they didn't have to bother your devs and your product managers and your team. And that's just a cool example of using the power of software, talking to software and.
ultimately making humans life easier by the fact that they didn't have to like, they didn't have to struggle. They didn't have to bother your other human developers. And I just think there's going to be a lot of examples like that in life where the more people, because because you understood like, Hey, my software is in claw in my GitHub. think you said repository and I can have Claude just like connect to that and like understand what's in there. And that, so for a non-technical person to be able to answer a technical question.
Adam Brotman (38:35.541)
Like that, that was just a really cool example. So I like that.
Benji Koltai (38:39.274)
Awesome. Really appreciate that summary.
Andy Sack (38:40.041)
Yeah, for me, I mean, I think a few things stood out. thought, know, Galley Solutions is just a great example of vertical AI company. And I really like, you know, vertical AI as the LLMs continue to expand horizontally.
and their reach just goes on and on and with an accelerated pace. I actually think really focusing in on a specific industry and a specific system of record data source, the recipes. I just think that you're well positioned to solve real problems in...
in food service. And I think that's going to be true in lots of industries. So I think that as a anecdotal example of a vertical AI solution, I Galley Solutions is great. And then I just want to thank you, Benji, for bringing up the Adam and I and the audience, the philosophical
existential fear that comes from adopting AI and really just the, you I thought you said it really well, which is where human beings and mandating the adoption of the technology is not nearly as effective as addressing some of these philosophical questions. So I think it's a great spot to end. I want to thank you, Benji, for being on the
the podcast and with that, thank you to the audience for listening to AI First with Adam and Andy.
Andy Sack (40:28.041)
For more resources on how to become AI First, you can visit our website, form3.com, download case studies, research briefings, and executive summaries, and join our email list. We also invite you to join our AI First community, a curated hub and network for leaders focusing AI hype into action.
We truly believe you can't over invest in AI learning onward.