Actually continued to do this since then is having contests off and on about who can do really cool things with AI. Let's go try to do this thing and or go try and everybody come up with something that's gonna be more valuable, bring it to the team, and we'll give out gift cards to whoever wins. We actually did a complete conversion in the team in Cinch in late February. We cut everything over in maybe early March, we cut everything over to ⁓ Claude and we were using the clo ⁓ desktop app, right? Once that came out and and you had Claude co-work, you had ⁓ Claude Code all all inside of one single app and we did the same type of thing. We basically told everybody build something that is going to make you more effective in your job and then come share it. And we actually voted on as a team then we had the leadership vote on like what it would be and and we picked you one based on how efficient that make it.

Feel like that that opening line for every one of them is this. Welcome everybody. You'll notice it's little bit different this time if you're watching, because ⁓ there's three boxes instead of two boxes on the screen. and ⁓ so we've got a special guest today. And so, you know, a lot of guests ⁓ or a lot of executives, they're still using ⁓ you know, AI as a chat box. Justin's not doing that. He's building where AI has to answer a phone, use real data, make right handoffs. ⁓ it needs to make good decisions and it needs to create real results. And that's that's just a very different ⁓ point of view of a of a practitioner versus a user. Yeah, and it it really raises ⁓ question ⁓ that you know a lot of other people aren't asking right now which is who maintains it? What happens when it doesn't work the way you think it's gonna do or you get a wrong output from AI? When so when can you trust the data? And then I think really critical nowadays because people are leaning in heavy in some cases. When do you build, buy, or just, you know, run a model internally yourself? I mean, all of those things are I mean, they're ROI questions. They're near and dear to me as a CFO because I want to make sure that I'm I'm spending the money the right way. Yeah. And so Justin Rae. Is the founder and CEO of Cinch. He's been working on the data gap between large companies and local businesses since long before like Gen AI was on the scene. So he's got ⁓ a long very long before. I mean, you're dating him. That's ⁓ that's pretty rough. My first guess right out of the gate. Look, man. You know, like who's got one person trying to be here is is is like all gray and ⁓ you know, and so you know, one of them's is an old man. But ⁓ Justin has been working for a long time on this and and I think you're gonna be able to give us a really great like perspective on how the the world has shifted over these last several years, especially. So Justin, thanks for joining us.

Good to be here. Yep.

Well yeah, so why don't you give us a little bit, you know, you know, as you talk about cinch, you know, and you know, give us the short version of the path of like, you know, maybe customer data to, you know, reminder marketing to the cinch that you're building today.

Yeah. Yeah. No, defin definitely. The the biggest, most valuable thing. I mean, we've heard for years and years data's the new oil, right? Those, you know, ten, fifteen years old type of ⁓ mantra. But but it really is still true, right? Is the fact that having good data and being able to manage that data in the right ways is becoming even more valuable than it was before, right? You know, you're not just connecting that to a BI tool and being able to get some insights out of what what's going to happen there, right? You're actually able to take action and work from that. And that's really one of the the core things that we've done at Cinch is we've really focused on how do we ingest as much data as we can and then match that up to a single source, which in our case we're focused on the customer. How do we know who the customer is? And then we know all these behaviors. We know what they're doing, how they're interacting with you, what is their relationship really like? And, you know, one of the things that we really want to do with that is like. How how does it turn into a virtual representation of the times when you could walk into a store and the store manager or the person in the store would know you, right? How do you turn that into that same type of b type of experience in the virtual world, right? Or in these digital or or the the AIH, right?

Yeah, which is where most of this is happening now too. I mean, face to face is ⁓ not it not nearly as common as it used to be.

Yeah, you know. We've got so many more people and so much more consumerism that we're trying to get out the door and how do we in interact with it a lot faster?

I guess for ⁓ you know, as as giving a little bit of you know, more background about, you know, kind of, you know, cinch and the work that you're doing. I mean, ⁓ you know, you mentioned the oil, you know, like this is the data's the oil and and that's where the value is and and ⁓ you know, how's that become how has that become more valuable as ⁓ you know, we're we're using different types and forms of AI?

Yeah, I mean just being able to access and and make decisions relevantly relevant to the data that's happening. Where at Cinch, we have for years used machine learning. Since our since our initial launch, we included a lot of different machine learning about how a customer interacted with a business. And a couple of those ways, number one is like customer valuable. How valuable is this customer to the rest of the ⁓ to the business? Relative to the rest of the ⁓ customers in that database, right? And so if that customer is moving up at the same trend, like most most people's ⁓ database, they they're engaging, they're becoming more valuable to their customers, and so their customers become more valuable slowly at a like a linear path, right? If a customer is not moving up with the rest of the path, then they're actually becoming less valuable, right? This this especially happens when You know, you've got pricing changes, or you just got, you know, over you know, if you've got a customer over the you know 10 year time period, right? Your pricing and what you offer 10 years ago is different than what it is today. And right, so that that monetary value has moved up and to the right. If a specific customer didn't and they stayed, maybe they're still spending the exact same amount, right? And you can't look at that and be like, ⁓ yeah, they're still the same value. No, they're not the same value because the rest of your customers have become more valuable. And so they become less valuable. And how do you look at that in a weighted way? That just the same way that an LLM applies weights to all the different words and processes that it's going in its decision making, you know, we're adding weights and measures to each of the interactions that a customer has, right? And over time, how are those changing? How are they getting better or worse? And so ⁓ so we're we've been putting things like that into the process. We have things like pricing sensitivity, which tell us. Hey, this customer is always using a discount and it's the deepest discount, or they never use the coupon anywhere, right? Those are the types of things that become really valuable at helping to dive in deeper to using that data. Like you said, it it it is it's the more valuable o oil now because you can get more value out of it. And so you can dynamically change a coupon on an offering because we know this person's ⁓ pricing sensitive, right? Maybe you even give them you know they're high value customer, they've always been high value, but they're always pricing sensitive. So if you give them that discount, they'll probably come in more frequently. Or you don't give them that discount, they come in less frequently but still spend the same amount. And so those are the types of details you can really start to get into now at the granular level and you can personalize that communication. in ways that were r really limited before, right? You still had to work in an aggregated level with most of your customer base.

Well and and I like I f it feels like that, right? The idea that we had to work in cohorts and and and and and we had to stratify and and and kind of build kind of right, we would A B test things and and try to figure out ⁓ a way to ⁓ to prove a hypothesis. And now it feels like you can micro test that and have a way better shot of getting it right because you're seeing a lot more trend, like you're saying, at that this really granular level. Like I know Jim, I know how you know, I know he likes to cut a coupon. ⁓

Yeah, f I win him.

You know. Yeah, I I I completely agree. I think, I mean, from what it sounds like to me though, ⁓ at least just what I know of cinch, I mean that A B testing now, it's it is happening, but it's happening rapidly. It's happening like basically around the clock, just to try to figure out what the next insight is and the next unlock is to to be able to sell the next widget.

Yeah. Yep. Exactly. I mean a a really cool use case of this recently is just being able to customize tones of an email, right? And so you still kind of work in a certain level of cohorts, but you can identify, you know, hey, what what tone does Jim need? Jim needs the educational one and Veer needs the urgency one, right? And and so changing the content of that email, changing the subject line to get you opened and engaged, right? Those are all done by A B testing and a machine basically trying to say, okay, well I'm gonna shoot over this one's gonna be the gym in a social proof tone, right? And so I mean from where we're going is we have about 10 different social proof or we not social proofs, but we have ⁓ about 10 different tones that we're trying to place customers into and identifying, okay, when you're trying to do this type of thing You need this tone. But when you do this type of thing, you need this tone.

I mean, just kinda going back to what you were saying earlier about how granular you can get things, I mean, seems like ⁓ eventually this'll be Jim has a tone, Veer has a tone.

Yeah, I think I think there's still gonna be certain cohorts, right? When you think about human and consumer behavior, right? People there are certain groups of people that tend to behave very similarly, right? ⁓ and so you'll probably get down to like get down to that, but you'll have a whole lot more, right? You'll your cohorts will be many, many times more. ⁓ but you'll still have certain groups of people that'll behave and interact ⁓ very similarly as far as that goes. But I think some of the other things that are really changing that are really cool are that You know, you can you can have the subject line and bait general content that is similar for one customer, but then you can get, you know, or you can have the tone as the same, but you have hyper personalization where, you know, maybe you're taking and this is again bringing all that data together, you're taking information from a phone call that you had with the customer or a text conversation and you're able to bring that in, right? Maybe maybe you're selling shoes and you had a conversation with somebody. Or they're chatting with a in a chat bot and they say, Yeah, I really love to run marathons. And I feel really lucky anytime I sell blue shoes, right? Or excuse me, I I I wear blue shoes. I'm gonna, I'm gonna really do great, right? Those are my lucky color, right? Now I've got some extra blue shoes in their size that are coming out, right? I can dynamically send those over and be like, Hey, here are your lucky blue shoes. Go buy ⁓ right? Those are the types of things that can become ⁓ you know, you still have like cohorted tones. But now you have hyper personalization based on a lot of the data that's coming in and that you have on this customer profile.

Well that, I mean, like that's exciting like that because I mean from what you were saying, you have hopefully at some level you've got like 10, 15, whatever it ends up being that are statistically relevant tones and ⁓ but then you can mix them right ⁓ down at this like micro segment level with like, but Veer likes blue shoes and ⁓ you know, he he you know, he he likes when the twenty percent off sale happens and ⁓ that is Yeah, that is powerful. One of the things that I want to ask you about is so you're doing all these really powerful things, but if you could possibly tell us, just give us a little glimpse into maybe what your stack looks like. Now I know there's some secret sauce in there. So whatever you feel comfortable talking about, but what is, you know, how do you as a practitioner go about implementing AI not only into the product that's consumer facing? But also into your back office and just into the DNA of your company.

Yeah. Well that's there's there's a lot to that, right? And it's it's not one of those that happens overnight. You know, we've been working with this stuff and trying it for a while and ⁓ and going going forward with that, right? I mean we as far as our software goes, we have a pretty typical software stack using ⁓ Postgres, using Snowflake, using React as as ⁓ general aspects of and and Python. in how the actual product works. But then we're able to connect in and bring in, you know, things like anthropic CPI and using the cloud, ⁓ cloud in that scenario. We have Gemini, we have ⁓ Codex or we're able to work with ⁓ Open AI as well as with their aspects and and ⁓ and then we have ⁓ local models. And so I think yeah it's really interesting because you have ⁓ that that's what happens product wise but as you kind of highlighted there's also the like operation side of things and how do we how do we function and work with those and you know one one of the cool things I think we started doing in 2022 we actually were even telling our team hey AI is going to be the change here right I want my team to know how to use AI and so we gave everybody the chance to use Chat GPT back then ⁓ telling them to try to come up with it. We had little contests and we've actually continued to do this since then is having contests off and on about who can do really cool things with AI. Like Let's go try to do this thing and or go try and everybody come up with something that's gonna be more valuable, bring it to the team, and we'll give out gift cards to whoever wins, right? ⁓ we actually did a complete conversion in the team or in in the cinch in late February. We cut everything over, it was maybe early March, we cut everything over to ⁓ Claude and we were using the cla ⁓ desktop app, right? Once that came out and And you had cloud co-work, you had c ⁓ cloud code all all inside of one single app, and we did the same type of thing. We basically told everybody build something that is going to make you more effective in your job and then come share it. And we actually voted on as a team, then we had a leadership vote on like what it would be, and and we picked who won based on how efficient that make it, right? So these are the types of things that while like the stack can be changing constantly because you know the world's changing so fast this way, right? But how do you culturally build the the opportunities for people to be able to do that, right? And that's what we've really focused on as far as that goes. ⁓ we haven't yet gone to the point where I'm willing to ⁓ buy everybody a spark box, ⁓ the NVIDIA spark and and put it on their desk. But but I think that if we did get to the point where where I knew that people were going to create the value from that, I wouldn't even be opposed to that, right? Is having that local model ⁓ to be able to run through and and just have something constantly monitoring where they don't have to worry about tokens. You know, might not be a a frontier model, but it's gonna be able to do some of these things that, you know, six months ago, honestly, you couldn't do with a frontier model. You can do on a local box now.

You know, I'd love to hear a little bit about some of the wins that you saw. And then I'd also like to know where you've seen maybe function by function, this whole use of AI spurring like creativity, efficiency, where have you really seen that take off? And where have been some of the places where you're like, man, I really wish. that function would have latched onto it a bit more. And y you know, 'cause I can imagine in your mind it's like, well, there's a priority stack. So you're paying closer attention to some functions than others.

Yeah. Yeah, no, I think on that priority stack for sure has been engineering, right? We've we've heard just all about how you can build so much faster and plug things together. And so definitely the R and D team, so our product and our engineering teams, being able to work out and get things going there. ⁓ did you have some of the beer?

No, I well, I was gonna say, like as you're kind of talking about like, I think ⁓ you know, part of it is, hey, you know, it sounds like folks are using it, you're getting it into the hands of end users, desktop workers, knowledge workers, engineers. ⁓ and ⁓ I think we all, I mean, like you we have talked before and we we we have wins. And I think an interesting way to maybe kind of like highlight is it's like what hasn't worked, right? Like what what what you know, like what what did you try? You're like, ⁓ Wish I could or, you know, like you know, like ⁓ 'cause I feel like that is where, you know, every s every you know, trip and stumble, right, is like ⁓ like, ⁓ I learned something and now I I'm you know, I'm do this better the next time and you know, trying to let folks know that even those of us that are using it all the time, we still trip and stumble every once in a while too.

Yeah. I think you actually hit it perfectly on the head how you said it, right? As we learn, we trip and stumble, and then we try something again, right? I mean, this is the you know, the the whole practice of everything, right? Test and test and repeat. And the aspect of how often are we giving context, what context are we giving to an AI, and then we're we're seeing what it does, and then we do it again, right? It's just a matter of iterations, right? And And you we were talking before a little bit about how sometimes we feel like we're getting behind because everybody else is moving with faster, right? But the truth is, is that's just us like everybody's getting these at bats, and the more at bats you get, the more experience you have, the more conversations you have, the more you get to learn from somebody else's. And so you gotta be having some both of these. But what's really fun about it is that's exactly how the AI works, right? And so if you're learning and you're the kind of person who really likes to get the at bat, even if you're gonna fail. And you're like, I know I'm I'm fine. If I fail on this one, I'm just testing, I'm just seeing what I can do, but I want to see what I could get out of it, right? And you get the result or you don't get the result, but you got the same thing with the AI. And you and that's what becomes really cool. Is is and especially as we start to think about how do we how do we replicate this or how do we help teach people this is how do we give it the context we need? How do we let it try something? How do we let it get the results of the the trial that it had and then go back to it, right? It's it's just a re rinse and repeat process.

But so a question for that is, and I've I've actually written about this a bit and ⁓ would love your take. What holds people back from using AI?

I I do think it's a little bit of like understanding what they can imagination. I I would probably say imagination. is you know, it ⁓ maybe even hope in a weird way, right? Is it's like I hope this can do it for me, right? But on the same side, because AI can have all its wonderful hallucinations, you also gotta like be able to have wins. So maybe it's the fact like Could it do this for me? Right. And I hope that it can, because I imagine it should be able to based on other things that I've done. I've seen it do this. ⁓ and I think that maybe that's just it goes back to iterations again. Like what of what do you do? Like what what maybe holds people back is the general aspect of of the other things that they've done or the way that they've used technology, right? ⁓ you know, it's l like we kind of talked about at the beginning, it's not just a ⁓ chat bot anymore, right? It it's It's getting work done. It's building, it's orchestration of multiple agents that have specific tasks of their own. And those tasks get done and and commingle with the other tasks of of other agents, right? And ⁓ and that stuff just builds on top of each other. ⁓ I think maybe you guys have heard of Hermes, ⁓ but ⁓ Hermes has their stages, their levels, right? And you can g hit the different levels of ⁓ of Hermes and and kind of ⁓ grow into those. And ⁓ I saw a really great YouTube video kind of explaining and walking through that. And I think that's the same thing for anybody. And if you can have a good experience at that base level, you can start to move up. Right. And it's how do you get people to have that good first experience? And I think that maybe going back to the long way of answering your question, Jim, is what holds people back is they haven't had that first good experience yet. that lands them in like the aspect of I can imagine what this could do now over here. Like I imagine more they can do. ⁓ there's a period of time where I feel like if people were using Chat GPT, they probably weren't having the best experiences that the people that were using Claude were. Right. I think that's changed today, but I think late last year that was definitely the case, right? Especially maybe even the first quarter of this year. And so you you're just like, ⁓ I can do this. Like, but I think that's how it needs to work, right? You gotta be able to Know that you can execute something and believe it's gonna be worth your time.

Yeah, we definitely want to hear from you. I I don't want to interject too much, but I think our my philosophy and yours is somewhat similar. although I didn't have the nuance on the first experience, which I think is a really like that's insightful. That's a takeaway that I'm going to steal from you. ⁓ but my whole thing is is I don't think it's fear. I don't really think that fear is holding everybody back. And I don't really think it's laziness. I don't think people are super lazy. I really don't. They're working hard all day long. Just to build on what you said, I think that having founders like you or business leaders or champions for this like you, allowing them the space to have the good first experience is what is holding a lot of people back. They can't get out of the hamster wheel of I've gotta get this report done. I've gotta this is for knowledge work. So I'm But maybe engineering is the same too. You guys are better to speak to that. But in knowledge work, it's like, well, I have to get the cash forecast out. I have to get the financial model done. I don't have time to sit and re engineer my entire process if the board needs this by tomorrow. That's my take on it. So I think you have actually solved this is one way to solve that issue is just giving your people space to learn.

Yeah. Yeah, it's definitely been really great. You ac kind of asked about some of those wins that people had had. And it was when we did that contest this earlier this year, ⁓ we we saw some really cool things, right? We saw our customer success team was focused on like, hey, how do we support our customers better? How do I how do I more efficiently manage my time to make sure that I have time to deliver really good value to the customers I'm trying to work with, right? Once you have people that were like building an agent that was going to Using cowork, go through all of their emails, go through the support tickets that came in, highlighting the customers that they've had, and highlighting you know what are the the issues that are gonna have a bigger impact and trying to prioritize those so they can tackle those faster. ⁓ they we had on our sales team something very similar in the aspect of looking at the email. How do I follow up with the prospects faster? How do I answer the questions and having an agent that's that's finding the answers for the ⁓ the sales rep. ⁓ we had one where the trying to do a demo in a way where you could mask all of the information that was on a page, right? And so you could basically we have like a demo that's built out for the quick lube or oil change space, right? And they he map made it so it could mask it to display pest control or it could display it, just basically we change it to make it look like the data was somebody else's, right? But there was still an underlying thing on top of that. We we had ⁓ like analyzing digital the digital ads and how some of those function. So just all of our teams are coming up with the things that like we're trying to solve their problem. And it was really cool to just to see that people were doing that. And ⁓ we'll probably be doing another one again here here soon, just ⁓ see how things have changed. ⁓ specifically trying to get people to try agent orchestration, right? How do you take two agents? And get one agent to do one thing and another agent to do another thing and then them to interact or an an a an agent to work off of the two of ⁓ right? And see what we can do do with our team. So it's a lot of fun. It really is.

That is really cool. Like I think, like to kind of tie like Jim kind of brought up the fact that like urgency, tyranny of the urgent, right, is is maybe something that holds people back. And I don't want to, you know, like like lose the thread where you know it feels like, you know, because what you've shared with us already is one, you've created space for your teams to be able, and you've created incentive, which I think is should be a takeaway for folks who listen, right? Is you know, by creating kind of Contests or, you know, that there is something tangible, right? To kind of work towards. I mean, ⁓ that that feels like it's not just one more straw being laid on my back of, you know, just something else that, you know, I I have to do, right? But this is something I hate to I hate to say it, but I think what you're describing is change management. And ⁓ you're gonna you're don't, don't. Yeah, I knew I knew we were heard that. I mean, I know that's consultant speak, but that is what we're we're dancing, we're dancing around. Like, how do you change an organization's culture or DNA to orient around this? Well, and it sounds like Justin, you're and I'm gonna make an assumption here and hopefully you can maybe like like you know, validate it. I mean, it you've created space, you've created incentive, and I'm gonna make this assumption that you you run your business with ⁓ a clear set of of metrics that people understand what is important. So as you said, they were working on projects that improved, but improved means, you know, like you know, something that that's really, you know, like like a hard number, right? I mean like something that's measurable, demonstrable, that ⁓ is is not just it feels new or it feels better or feels faster, but but it feels like maybe you're like it so the assumption is is that your your team, you've also created what the finish line You know, like what w you know, like the North Star. Everybody knows where that that that star is.

Yep. Yeah, exactly. I mean, it's just like if you you go back to that loop of AI, right, you gotta know what the end result, what was the mission I set out to do, and how do I know when I hit success, right? What is my success criteria? What did I try? How did I end up with that? How can I try it better? Try it again, right? Just keep going around with it. And yeah, I think, you know, kind of tying this back to ⁓ some of the previous part of the conversation, but where You know, we we did have a period of time where we s one of our teams, they ⁓ they're they're everybody's just crunching at cinch. We're just all trying to get everything done. There's a lot going on. We got a lot on everybody's plate, and we had one of our teams that's just like, you know, asking us to try this thing in AI feels like you've just been another thing on our plate right now. This to to Jim's point, right? Now that team They actually knew the value of AI. They knew what v AI could do already. This wasn't like a you know, we were we were jumping into something totally new. But what I didn't hear right at the moment from them was it's like, look, we have, we gave it that try and like we didn't get like we didn't get the results that we wanted, right? And so going back to that, like having a good first experience, they knew what AI could do. This was something we were trying to build and put together and it just wasn't it wasn't helping them. About Five months later, we totally changed up some things in from product and we were coming to that team. We were talking to them to try to see what we could do and totally different experience. Right. ⁓ now I think if you took that exact same thing away from from them, they would be like, I don't even like you you want me to go back to working that hard? Like you just gave me a tool that made it a lot easier, right? I often think about the the differences in these tools, like. We can all visualize the difference of a hammer and nails and a pneumatic ⁓ nail gun, right? And how much faster you can just go through when you're putting a roofing project down or anything else, like nail, nail, nail, nail, right? As opposed to like, I gotta hit this a million times with a hammer, right? Or ⁓ in the automotive space, right? You got a pneumatic wrench and you're trying to put on ⁓ lug nuts back onto a tire, right? Well it's gonna take me a lot of wrenching on that without that. But I get this pneumatic wrench and I can just this is this is Right. And ⁓ and that's the way it is really with AI, right? Is now we're kind of getting these real power tools. It's not just like, hey, this was a thing that helped me with like as a knowledge worker to get things done, right? Excel has been the tool forever, right? Excel used to be one of my favorite tools. I use Cloud way more than I use Excel now. But I used I used to have fifty Excel files open at any given time to help me manage everything and where it was going. And ⁓ and so I think Yeah, going back to like having that good first experience. How do you get somebody to have that experience that lets them know like, ⁓ I can have some trust, I can move forward with this, and it's gonna actually be worth my time. It it it saves that change management fear, right? It it gets them into, ⁓ I actually want to change because you know, I can see that I can change, you know, ten tires in the time it takes me to change one tire, and or I can you know, put in fifty nails and the time it takes me to put in five. So y y you're showing that ten X experience in in functioning and getting things done.