Future of Freight: Automation, AI, and Industry Transformation with Levity Co-Founder, Thilo Huellmann | Episode 233

Speaker 1: 0:19

All right, welcome back for another exciting episode of the Freight 360 podcast. We've got a special guest today. It's going to be a great episode. We'll get to him in just a second.

Speaker 1: 0:29

If you are brand new, you caught us on a great episode number 233 here, but there's a ton of other content and episodes. You can check us out wherever you listen to your podcast draw on YouTube or on our website, freight360.net. Share this with your friends, leave us a comment, leave us a review, click the subscribe button all the good stuff that helps us. Share all this great content with anybody out there that is in Freight Brokerage or considering getting into it. If you'd like to learn more about us and our training opportunities, check out the Freight Broker basics course on our website. It's a full-length, self-paced online course that will show you everything you need to know to get your Freight Brokerage started and off the ground, get those customers, build the Care Network and even how to hire the right employees when you get to that point in time.

Speaker 1: 1:16

But without further ado, we've got a special guest, tilo. He is the co-founder of Levity, a recent new partnership, and sponsor that we've added to the Freight360 podcast lineup, and we're excited to have him on the show today. Tilo, how are you, man? It's good to have you here.

Speaker 2: 1:34

Yeah, thank you for having me. I'm doing well. How was it Awesome?

Speaker 1: 1:38

So, for anyone, we'll get into Levity in quite detail in a bit here and the problem that it's really helping solve. But for anyone who doesn't know you or isn't familiar with you or your background, give us a little bit of a recap on who you are, what your role is and how you got into this Freight space.

Speaker 2: 1:57

Yeah, sure, so I'm one of the founders of Levity and I'm the technical co-founder and CTO right now. I also do a bunch of other things, as many founders do, and, yeah, in the end, what we do, we help Freight brokers to deal with email in a more efficient way. So anything that is repetitive takes up a lot of time. Copy-pasting information from emails into other systems. These kinds of things are the things that we can help with and automate so that you can really focus on the more valid tasks, like building shipper relationships, building carrier relationships.

Speaker 2: 2:37

Yeah, the reason why we got into this is a little bit accidental, a little bit former startup experience at a different company, but happy to jump into the details in just a bit.

Speaker 1: 2:51

Yeah, for sure. Well, it's good to have you on, Ben. How are you doing today, man? Doing well.

Speaker 3: 2:56

Is it warmed up at all? In Florida it has. A little bit Cold snap was over, at the very least. It was only about 60 this morning instead of low 50s and high 40s.

Speaker 1: 3:05

Nice. It's really strange in Western New York. Here we're in a really weird transition phase between winter and spring right now. So we had snow a week and a half ago and then I was out of town and I got back and it was 50, 60 degrees. It was like 65 degrees yesterday and it's going to be 20 degrees tonight. So we're in that very bipolar weather phase, as we aren't really sure if we're still in winter or if the Punx-Otani fill was right on Groundhog Day and we got an early spring. But it's looking like after this cold front comes through we'll have a nice warm weekend and a week or so, but I'm looking forward to the way for spring and summer. Yeah, what do we got for sports, right? I mean, baseball season is around the corner. We got spring training getting into it, the match, right I know it's texting with you about it the other night and nighttime golf under the lights. That's kind of cool.

Speaker 3: 4:03

Yeah, it was. I think Rory ended up winning. I didn't stay up for the very end I was checking it again actually but apparently it went into overtime. There was a tie, they tied the last four holes and Rory ended up hitting it closest to the pin and like a playoff thing. But overall it's a pretty cool thing to watch. I mean, I don't know, it was local to me, it was only a couple of miles up the road and I didn't know this. But Lexi Thompson's, also from Delray, which is like literally right next door to me.

Speaker 1: 4:27

Is she one of the girls?

Speaker 3: 4:28

that was playing Is and.

Speaker 1: 4:30

I looked up after.

Speaker 3: 4:31

I don't watch too much LPGA, but she is the youngest person to ever win an LPGA event. She won at like 15. And I think she won a major at like 16 or something like really, really young. I mean, just had a crazy career and is still playing very, very well, but nice, yeah, it was cool. To be honest, what did you think? Compared to like I think the one with Brady and Steph Curry was just more entertaining to watch. To be honest, there was just a little more shit talking and it seemed more like playing golf with my buddies. This seemed to be a little more yeah up kind of corporate.

Speaker 1: 5:05

And again, it was still an interesting Monday night to watch, I thought it was weird for me and maybe it was the broadcast, but I was watching. I think it was on. Tnt was like the what I had it on for Verizon. It was so hard to hear the players. Like I get here, charles Barkley, just fine, and then I, these guys that are wearing AirPods, I'm like like leaning into my TV, like what? Yeah, can't really hear you. But yeah, the cool thing was under the lights. We've got a local three par three course near me that does that. You can golf till midnight. I I've played global. We do that. We have like a glow in the ball Up the door.

Speaker 3: 5:42

It's funny you said that like because you texted me that during this and I'm like I've lived here for like nine years there's no place. There's a couple in Miami where you can go at night and there's some miniature golf up north at night Like Tiger Woods. This place just opened up but there's nowhere to practice and it gets so dark so early here that I'm like I would love to have what you have up north. But up north I used to do night par three and we used to play go global, like once a year. It was like a charity tournament and it was just cart girls, beer and yeah and you just seem a little dangerous.

Speaker 3: 6:15

You just see golf balls flying to there, because you don't see that during the day how close golf balls really are to people, but at night that's all you see is just streaking golf balls past people's heads For sure.

Speaker 1: 6:28

Teal. I got to ask you what's your go to sport? You got a big fan of anything.

Speaker 2: 6:34

Well, I used to play basketball in high school, but not so much these days, so mostly running, I would say.

Speaker 1: 6:41

Okay, that's awesome. It's funny. I went for a run yesterday with my wife and she's pregnant, so like she's six months pregnant, running around with a baby in her belly and I'm in my later 30s. I'm not in nearly the kind of shape that I was 10 years ago when I used to be a big runner and I was like man, three miles was like we did a 5k yesterday. I'm like that's that feels like I just ran 10 miles and I don't have it in me anymore.

Speaker 3: 7:08

But anyway, I'll make you feel a little better, Nate, Like when I ride my bike down. I ride it on A1A and I've been going more lately. It is so depressing to me when I'm like doing better on my time. So like I look at what I'm doing, I'm trying to improve and just five elderly people just smoke Like, pass me like I'm standing still and it's like the most humbling thing ever and I'm just like, yeah, every time it happens to me it makes me a little bit sadder.

Speaker 1: 7:35

So what do we got for this news article here? This is a oh yeah, ikea convoy. Is this the one that we had in the newsletter yesterday? It's a big story, though, so the I got to pull this up, and then we can kind of give our take on it. T-lil probably has no idea what we're talking about, so there's a big news piece that's been going around Ikea, the furniture company. Right, you can buy everything in a box and a warehouse and take it home and assemble it, and they are suing convoy. Was it for what was the? What was the reasoning for it, though? Is it something with carriers not getting paid, or something like that? And then they're trying to come back around to Ikea.

Speaker 3: 8:23

I am looking into it now, to be honest. I mean it's basically saying at the top that they disclosed basically all their financial statements, but I'm not seeing what the original lawsuit was related to. I think it's convoy not paying the carriers for free.

Speaker 1: 8:41

That's what I'm thinking.

Speaker 3: 8:43

Because it says yeah, like clearly.

Speaker 1: 8:45

I'll just read this from our newsletter. I'll read this to you. This one out yesterday. Today is Wednesday.

Speaker 1: 8:54

Ikea has filed a lawsuit against convoy and 58 other parties, including carriers and freight factoring companies, in the wake of convoy's $3.8 billion collapse in October of 2023. The lawsuit seeking to determine who should be paid from an amount totaling $519,000 roughly, was initiated due to uncertainties following convoy's financial downfall and the subsequent acquisition of its tech by Flexport. Convoy, which has not declared bankruptcy, failed to pay carriers for loads delivering for IKEA. That is your main answer Leading to a contractual dispute. Since IKEA's agreement with convoy stipulates payment only upon convoy's fulfilling all obligations, including paying subcontractors, ikea has deposited the disputed amount with the court seeking a resolution to prevent further claims. Despite the collapse and sale of its tech assets, convoy's liabilities remain unsettled, with ongoing concerns among carriers about unpaid dues, contrasting with the positive reception of convoy's tech under Flexport's management.

Speaker 1: 10:03

So interesting, stephen added in here. This article came out after the newsletter and talks about the rates between IKEA and convoy. It was to add on the conversation with rate transparency. Okay, so Stephen put that. Our producer put that link in the show notes, which is why Ben and I were trying to play catch up. But yeah, there's your story. I've seen it on various different media outlets, from like freight waves to like CNBC. So big deal when you've got numbers that big, that's pretty loud though.

Speaker 3: 10:36

Yeah, here's what gets me, though, right. Like I do feel like it's a snapshot to try to justify a larger context, and, like, again, if you're going to look at contract rates for a large, large carrier working with a large shipper and you're going to pick October 2023 to show the disparity, it's going to be enormous, right. And if you look at that number a year and a half before that, it's going to look the opposite direction. Yep and for sure. Again, I do want to point out to anybody that's looking at these things right, like, again, this is a snapshot window in time. This is not over the course of market terms.

Speaker 3: 11:13

Right, and again, a shipper needs to negotiate because they need to make sure their trucks are picking up on all their lanes when they're all supposed to. They will pay a premium for that, and sometimes a broker is making nothing, or 12%, when the market shifts. That contract rate stays there all year. If it ships in the other direction, which it did in 2020, all of these were upside down and they likely lost money, right. So, again, I do feel like these become what are they called? Like a red herring to justify a larger point, when it really doesn't at all is my big takeaway, but I'm curious to see how it plays out. I'm sure there'll be lots of people up in arms and using this to justify why all these things need to change. And the reality is this isn't really evidence of much other than what the market does in a loose carrier market, with a large brochure For sure For sure.

Speaker 1: 12:02

So we're going to get to our topic today and talk with Tilo. So, tilo, we got a chance to meet you I don't know earlier this year and when I saw what Levity is and what it does, I was kind of fascinated. I'm going to let you tell the story about that. But what I found interesting and I won't reveal too much, but there's a I had met a company in the past that did staffing, like they're kind of like a staffing firm, and they had their own like internal proprietary thing that helped people out with some sort of email automation. Because the reality is, whether it's a quote request or a load tender, email is like what totally dominates the way that we communicate in our industry. Right, ben, you got something to happen with.

Speaker 3: 12:48

Yeah, because I was thinking about this when you were doing the intro and I was going to point it out, but I waited until now. Is it like? I mean, to be honest, to keep it simple, it seems like he's solving the reasons why I need two giant monitors to be a freight broker. Right, because I'm constantly an email and one and my TMS and the other, and half the time I'm transposing info from one to the other or using a notepad in between, because my TMS switches to the next screen. I got to go back up and then see another load, open that attachment and transpose it over here. Like there is a lot of time for every freight broker operating between these two tools, right, email and TMS I mean, it's where most of our time goes other than talking to people.

Speaker 1: 13:28

So I mean we'll start right there. Tilo, talk us through. How did levity come to fruition? What give us kind of like the founding story? I mean, you're a co-founder of the company, so you've been around since day one. Tell us a little bit about how I mean you mentioned earlier on too, kind of by accident right, and that's kind of how it's. A lot of products and things get developed as not on purpose, but you kind of stumble across some that you didn't go looking for and it gives you an idea.

Speaker 2: 13:56

What does that look like? Sure, yeah, I mean, it's a long story and definitely a wild. Back by now, I think my co-founder and I we got started over five years ago and back then we were actually living together. He was working for a fairly young startup back then, and me as well. My startup that I work for doesn't exist anymore.

Speaker 2: 14:23

This is one of the largest freight forwarders in Germany now, so back then they were called Freight Hub and he was, I think, the sixth employee or so. Now they're called 4TO, kind of like the German Flag Sport Okay, that's how they look at themselves and he was having a problem with a really, really repetitive and time-consuming process for them. It wasn't actually about email in that case, so I think it was a spreadsheet of potential customers that were just starting out, and what they were trying to do is they were going through the websites of each of those online shops it was right, the ones that import goods from China to Europe and they wanted to distinguish between online shops that do that and online shops that don't, and the only way they could find was to go on the website, look at the photos of the products that they were selling and having interns and working students do that kind of job.

Speaker 2: 15:31

Very manual process, yeah, and then going back to the spreadsheet being like yeah, I looked at the photos, it's probably somewhat important something from China.

Speaker 2: 15:41

And I had something similar in a very different industry real estate where I was working in a startup and they were basically doing the same and I was talking to him and we were thinking, okay, this can be the right approach, why is this not automated?

Speaker 2: 15:58

And we realized that it can be automated, but the tooling isn't there, the internal knowledge isn't there, the people aren't there that can do this, and we thought, okay, what can we do about it? And then, for I would say at least a year, a year and a half was just the two of us in our living room figuring out what kind of product we built to solve this kind of problem. And then eventually, we also incorporated, hired the first engineer and got the ball rolling. That was around 2020 when COVID hit. And then it was a really, really long time of product development, figuring out who's the right customer, what is the right use case, and eventually, after a full round trip, we went back to freight and particularly intermediaries that are between shippers and carriers that have to deal with the most amount of emails.

Speaker 1: 16:58

Yeah, we're dealing with both sides, right? You know, not just a shipper to a carrier, it's shipper inbound to the, also as a broker, then the broker outbound to the carrier, right? So for sure, it's double.

Speaker 2: 17:10

Exactly and we weren't even going after that particular segment. But even without going deliberately after it, it became the most prominent in our customer base and when we saw that we were like, okay, interesting, these seem to be the most engaged, happiest you know, customers that spend the most with us. So let's really double down on that. And that was around summer last year when we realized that. And then a few months later I flew to Chattanooga to freight waves. Now GEO has been at Manifest. We made some really great connections there and we said that with growth potential. So that's why we started engaging with the community more and really putting it out there.

Speaker 1: 17:57

So what are? Can you give some specific situations or scenarios where the product actually applies, like I mean, like talk to me, take me through. Like, hey, customer sends an email, take us through that process and what levity is doing and how that takes workload off someone's plate?

Speaker 2: 18:16

Sure, maybe let's start with the shipper side of things. When you know a new potential load is coming in and you know, as you all know, right in the spot market, there's a lot of stuff going on. Not everything is something that you want to take, right, and a lot of stuff is also coming via email. So the shipper might want to ship something from, you know, chicago to Dallas and is sending that to a bunch of brokers that he or she trusts. Right, sending out those emails asking for a fast reply. And, to be frank, a lot of shippers, they just take whatever reasonable offer they get back in the fastest amount of time, right? So, yeah, speed is of the essence there.

Speaker 2: 19:09

So what does the broker do? Right, they read the email first of all. They need to understand oh, this is really important right now. I need to quote this if I want it. Right, then I need to get all of those details out of the email into my rating engine or however I determine my rate. I need to prepare quotes and send it back, and then I'm done with that task for now. Right, and this can take minutes, right, and maybe I don't catch it right away because I get 100 emails or I just started my day and my inbox is full. What levity does is it directly connects to your inbox. So we're not a tool that you use where you have an interface that you're interacting with. We're operating in the background, where you just use your tools like you do your output, your Gmail, your front, whatever we connected it immediately in real time understand

Speaker 2: 20:03

okay, this is a new request for a quote, right, and we get the right information out, we format it in the way that is needed for the TMS that you're using and we're also pushing the data there.

Speaker 2: 20:18

We're pushing it to the rating engine, we're pushing it to the TMS, we're pushing it to any other system that you might have set up. And then also, what happens is that you can define any kind of business logic so it's very flexible where you can say you know, if the load is below 500 bucks, I'm not taking it right, so you can immediately put in some rules there. And then also, on the carrier communication side, right, if you are doing the email last and you're asking for bids, right, or you're sending a tender and you're asking your carriers to either accept or decline, you have the same issue, right. They coming back to you, they might want to renegotiate, they might just say, okay, I'm accepting this, but then again you have to you know account for that, go back to your system and put that in. So we're kind of tackling both sides and we're living in between those systems in the background and taking care of these things.

Speaker 1: 21:23

That's interesting and you've got a good point too right Timeless of the essence in a lot of this stuff and Ben can probably attest to this as well that there are. There's plenty of customers that will send out a load list and it'll have the price that they're going to pay. It's non-negotiable and it comes down to who can get me a truck first or you know something along those lines. But I've had plenty of times where you know a individual rate request goes out and you could see, you know you're one of like 12 people on the CC line or maybe 50 on a blind carbon copy and if you wait 30 minutes before you respond, that load's probably already booked by somebody else. And you mentioned Tilo, a couple of things that are like super normal, right Beginning of the day, beginning of the week. Maybe I'm busy and I think having that in the background to just kind of do it for you is fascinating.

Speaker 3: 22:20

Can I? I would like to go through an example too of some of the parameters you can put in there. Right, so can you walk us through just that first example? Right, customers send you a load right and says, hey, what do you need for this right Jump ball freight that we kind of talk about? First person to get it wins reasonable rate. They got to get this covered and move on. That's the shippers point of view, right? So speed's super important to us. How does levity and what are some examples of the variables we can use and how that relates to a rate? Okay, so, like it goes to the TMS, it checks our average rate, it can send them a margin back. Like, how does that process work again, like kind of, as Nate says, barney style, step by step? Can you walk us through?

Speaker 2: 23:04

Yeah, sure, I mean, it depends a little bit on how you, how your rating works right, how you want to price things, and then of course you know whether it's truckload or LTL, or, you know, domestic or cross order, all these different variables. So we've seen different setups depending on that. Some customers use sonar to determine the rate, some, you know, do it via DAT, others are already green screens customers, others have spreadsheets right. So it depends.

Speaker 3: 23:38

But the way, let's go through an easy one. Yeah, full truckload, full truckload. Right, yeah, it's just going from whatever Stockton to New Jersey. We want a 10% margin on the DAT average. How would that play out? Customer sends an email. Then what happens?

Speaker 2: 23:55

Yeah, customer sends an email. First of all, we detect that it is that type of request. Right, we know that. Okay, it's that. Once we know it's that, we have a schema setup and that's also something that the customer would say. They would say I want exactly these points of information. You know origin, destination, you know equipment type, all the details that matter to determine the rate. Then this gets pulled out by the in the background automatically and sent to where this, where the rate is determined.

Speaker 2: 24:36

So could be greenskins, for example, where they have an API that takes in these inputs and give a rate plus, you know arrow margin, back what they think is fair right now and then, based on that, a reply is drafted. So, first of all, it's not sent out right away by default. You can also do that, but usually what we do is we prepare the reply and then there is a final step where the human has a second look, just a sanity check, to see okay, it has a lot of these steps. It has identified the request, has gotten the details out, it has used those details to determine a good rate. Maybe I want to add another hundred bucks. I can quickly do that right and send it out.

Speaker 1: 25:27

Right, that's pretty good, because I always think about, with anything, whether it's like machine learning or AI or automation Like I just think about what if, let's say, there's an issue in the customer's email or on green screen, like you know, somewhere along the line and it comes back and it's like oh no, I just accidentally offered to move this at half price. I'm gonna take a huge hit, right. So I'm glad to hear that it gives you like the final blessing for the email to go out, because maybe that could be a problem for sure.

Speaker 2: 25:59

Exactly. Yeah, that could be pretty bad. And of course, you can also put in a bunch of rules and you know checks and balances where you just see, okay, it doesn't make sense for a load to be, you know, 50,000 bucks, it doesn't make sense for a load to be 10 bucks, right, these kinds of checks are definitely there. But, yeah, we also saw that having a little bit of human in those critical moments is definitely important, but nobody needs a human to copy paste stuff from an email, right?

Speaker 3: 26:31

I have a question too about that. So if, let's say, your customer, right, sends an email and it's fine and it's standardized, and this is functioning, and they send something that's a little ambiguous, right, like I'm thinking a city that is located in multiple states, and they just omit the state because they know you're used to seeing that that's where their facility is. So it's just. I wish I had an example. But you get the idea right, like in that scenario, like will the system ask me, like in that step, hey, is this so-and-so California or so-and-so Massachusetts, for instance? And then you're able to verify that. And then does it like I kind of learn that into the future. How does that kind of? Yeah, out of curiosity.

Speaker 2: 27:11

A really, really good question. Yeah, there are a lot of cases where either you know there's implicit knowledge that you know the shipper and the broker have, where they know what it means. But an objective system that doesn't have that implicit context would not know, right. And also there are cases where shippers just send stuff that is hard to interpret, where even anybody would ask clarification questions, right. So in both cases you can again set up a rule where you say, okay, what happens if information is missing or if something is not clear, where I cannot determine the rate either, you know it gets rather to human and the human is told, hey, I have all the information, but I'm not quite sure about this one data point. Can you check, maybe? Or you say, well, if the system is very sure that an important data point is completely missing from the email, you can also send. Well, exactly, immediately send out an email back and ask me hey, is this what?

Speaker 1: 28:21

it could be. Yeah, it doesn't say if it's flat bad, reefer van or something like that. Yeah, Interesting. And Ben, I got an exam. I looked it up. Washington there are 91 Washington's in the United States.

Speaker 3: 28:34

Springfield I should have remembered to from just the Simpsons. I know is intentionally ambiguous because there's a bunch of them in other states too. But yeah, washington, springfield, lincoln, I'll bet, is another one too.

Speaker 1: 28:46

Franklin, clinton, madison, arlington, common ones Anyway.

Speaker 3: 28:56

Can you take us from here, like from this step? So that's awesome. I mean, that saves tremendous time for quoting. The other thing I was thinking is a use case, right, is it? That's great for bringing on newer brokers too. Like, if you can kind of set this up for newer brokers, it's giving them an idea of where it should be. They can go and verify it. It's kind of spoon feeding them to let them know what should be done at the speed it should, and it's going to train their behavior too, that this is the responsiveness required for this type of business, right.

Speaker 1: 29:24

Does it read attachments? I wanted to. I know Stephen had put this in the chat. Does it read attachments or is it just the text?

Speaker 2: 29:30

Yeah, it's also working with attachments, even if they're not, even if they're scanned documents, which is sometimes, you know, a little bit more more difficult. We can also work with that.

Speaker 3: 29:43

So from there, right like building a load, say they accept my rate, it goes back to them, and they go okay, tender the load. And that stuff comes right back through and an attachment. How does levity deal or assist with that scenario?

Speaker 2: 30:00

So if it's being booked right, the process is very similar, just that we probably get a little bit more info right.

Speaker 2: 30:10

It's more info needed to to continue exactly all the details you know, nodes or whatever, and it's required in the, in the TMS, and again, there then have connectors to different transfer management systems. Could be, you know, an Alice, could be Revenova, tovo. Yeah, in many cases we also see proprietary systems, to be honest, and yeah, again, the email comes in, it gets detected as okay, we want to book this load. Great, what do we need, right, which information do we need to pull out? Also, there's a normalization layer, because you know, some people might write usd, others might use the dollar sign, and then you have different formats and sometimes languages you know there's a lot of stuff that can go wrong, and not only you know, everybody formats their emails slightly differently.

Speaker 2: 31:17

But that normalization layer is also taking place. And then, once it's normalized, we can then push it to the respective TMS, and again, what we also then see is there might be another human checkup. So in some cases you can create a draft, also the TMS, something that is not actually booked yet, but all has all of the details copied over, and then someone else can double check. Does this look good? Okay, great, let's confirm it and that's it basically.

Speaker 3: 31:49

That's the most time-consuming thing I feel like of being a freight broker, like those two things I spend more time on Than anything else I've probably ever done, except for prospecting. I mean Just and again, like I had to do this a couple days ago to put like 20 some containers in. All of the lanes were the same, just the numbers were different. But it literally took me about 40 minutes to do something that simple, just to make sure everyone was correct. They were all lined up and again they're in attachments. So I got to open each attachment and I got a look at the screens and cross-reference.

Speaker 3: 32:20

And again, this is very remedial work right. That absolutely I would rather have someone else doing right, but it's super time-consuming, it has to be done. And it's also one of those things that, like 99% correct is a failure, like if it's not a hundred percent correct, the load doesn't go, nobody picks it up and then somebody calls you in an hour and you got to spend 45 minutes figuring out what went wrong, correcting this air, when all of this could have been prevented Just by having it accurately transposed without a human counting out 13 digits, right, let me ask this Tilo is there.

Speaker 1: 32:58

so I mean, levity is not just a concept, this is a real product. You guys have real customers. What are you seeing for Like? I'm curious on, like, reduction in time for a process, whether it's a call, request, a track and trace, a Load being built, something like that? What are you seeing for, like, how much time is being saved for one of these tasks versus, hey, it used to take us 15 minutes and now it's whatever?

Speaker 2: 33:25

Yeah, it depends a lot on how well Set up someone was before and how good the people are. Of course, that's true.

Speaker 1: 33:33

If they're an absolute disaster, it's probably gonna save them a lot more time, right?

Speaker 2: 33:37

Yeah, exactly, but Well, what we've seen, I would say as an average, is that when we talk to customers that have been live with us for a while and that, see, you know how it affects their day-to-day In terms of spending more time on revenue generating activities, right, they can Usually save something between two to four hours if they have been doing a lot of this stuff, right?

Speaker 1: 34:06

So In a day in a day.

Speaker 2: 34:09

Yeah, depending on what their job is right. Yeah, we have. We have one customer where they had a full team taking care of these things so that other people don't have to do it, and we were just automating a bunch of these things. So they had a. They had an email triage team. That the only thing that it was reading the inbox and, you know, making sure that everything goes to the right place and Copy-pasting information. So in their case, it really saved that team a lot of time.

Speaker 1: 34:43

Wow, I think what's the same thing.

Speaker 1: 34:45

This is not. This is not replacing humans per se. It's more so offloading the mundane, like tedious, repetitive Data entry type tasks that nobody likes to do. Right, and I like what you said. Like it gives you more time to produce today. Focus on producing new profits. Right, you can focus on those revenue producing activities instead of going through. So, like there's people out there that are like me and that are different vastly, but when it comes to me and email, I am like very anal about my inbox. Everything gets read, everything either gets deleted, responded to, filed somewhere or whatever, and there's times that I spend an hour or an hour and a half in a morning Just getting what I would consider caught up on email and I didn't, you know, none of that necessarily led to anything revenue producing.

Speaker 3: 35:34

So I Was thinking the same thing you said, nate. Right, like and here's that, here is the use case. That really gave me an idea Of how much time this would save. Right again, I do lots of containers, so I do a lot of this type of work and it's very repetitive, even when it's slow. You do a lot of this and the difference between the new outlook and the old version of outlook is the new one.

Speaker 3: 35:56

You can search by PDFs and again, just that little difference, right, saves me hours a week by being able to go because, to your point, like they're scanned, so like you really got to look at them and they're not always legible and sometimes there's an error. And like, just using the new outlook Versus the old one saves me sometimes hours a day, definitely hours a week. And like this is that almost like on steroids, to where it's not just the search function to find the PDF quicker, right, which saves me a ton of time. Like this is literally doing the most time-consuming step for me. Like when I was at TQO and did a lot of content, we, like three people, just did this all day, just making sure the container numbers were correct, that they were organized in the TMS, that the drivers had the right ones and just doing this All day and then correcting the invoices after the fact, like it's just it's super tedious work and it's super time-consuming.

Speaker 1: 36:50

Yeah no 100%. Um, steven put a question in here. So it says levity can read documents. Has levity found a use case for the accounting functions like printing documents to loads, verifying? Okay, so is that you know outside of what it's doing with, like rate requests, order entry, track and trace? Is there any other functionality? Or is there? Do you guys have a vision of, hey, we're gonna actually be able to do stuff on the billing side, etc. What is? How far does it the reach go?

Speaker 2: 37:24

Yeah, I mean Theoretically. Yes, we can do that. Right now, we really focused on more of the, you know, critical things of quoting, load building and so forth. But also Afterwards, when everything is is done. Right, there's a lot of paperwork that has to has to be done, and it follows the same pattern. Right, you have some unstructured data, for example a Bill of lading or something like that. You have to Get Information out, you have to reformat it, you have to put it somewhere else. So I Mean, eventually we will also build these accounting integrations. Right now, we've mainly focused on TMS rating and that sort of stuff, but I can see that Also expanding to that you know what's cool is like.

Speaker 1: 38:13

So Ben and I have both like we're big fans of a lot of like the generative text, right, whether it's your chat, gpt or Gemini, which I think was barred and then what's? The Bing is a co-pilot. Anyway, it kind of reminds me of that. We're like you know, it'll take whatever you give it. It can go do some research and then it could come back and present something to you. It's kind of like. It's kind of like that, but for free brokers, right. So is there is there any correlation between Kind of this AI boom and levity coming to market in the recent years as a that? Is it based off of like technology that didn't exist and its current state back then, or is it or is it just totally random that the correlation and timelines?

Speaker 2: 39:00

No, it's definitely not random.

Speaker 2: 39:04

Yeah, I mean, when we got started, that was around the time where the first Language models of that sort, which are now powering chat, gpt and you know all of these you know recent developments that we've seen are all powered by the same type of AI model.

Speaker 2: 39:22

In the end, the only the only difference and why it has exploded so much is that the AI researchers have realized that making the models bigger and feeding them with more data and training them for longer on even bigger and bigger supercomputers just leads to an explosion in capability.

Speaker 2: 39:42

And the difference for us when we got started is these models were Working the same but were way less powerful, and it was much harder to get a customer Started because they had to have a lot of training data from their own Historical data, providing it to us, then us using that data to train a model for them so they can could start using it. Nowadays, these models are so powerful that you can just Tell them what to do right and they they will do it for you, basically, and that has made our product development also so much easier, because we were always about applying it to actual business use cases and automating repetitive workflows and that sort of thing. But we were Not the AI researchers or you know people that you know Inventing new kinds of models. So that really helped us to all our customers faster Get more customers to the point where they had a really performing well performing system.

Speaker 2: 40:48

So it's not 80% but the 99.9% correct right, what we need, and it does so with very little Preparation these days. So this is a big correlation for us, where everything just got a lot easier for us.

Speaker 1: 41:08

I find this whole AI boom Fascinating, like I love to follow the news and all the developments and new releases and I'm curious. I mean I think we're just kind of at the beginning of when this is gonna take, where our you know where, how tech and tools impact businesses, especially in our industry. What is, what's the vision for levity Is there? Is there a grand like hey, we're not gonna be just on these core Tasks, we have this other. You know, eventually we want to be able to do something else within brokerages. What's that vision look like for the company, for the product?

Speaker 2: 41:48

Yeah, I would say the product we have built can already do a lot more than what we just talked about at the core. It's built in a way where it's very modular and flexible and, you know, the quoting process is just one flavor of how you can use it in the end by saying my data is coming from Gmail, I want to read Subject, body and attachments, I want to get this data out, out of this block of data, right, I want to format it that way, I want to push it here, right? So that's pretty generalistic. I would say we are sort of Packaging it that way right now where we address freight brokerages 3 pl, it's by just telling them hey, you can do this with email. So they have kind of an entry way into how this technology works.

Speaker 2: 42:46

But it's extremely versatile and customizable. So in the end, that's one part and the other part is Oftentimes the more powerful, the more versatile you make something, the more Savvy the user has to be. Right, it's often leading to learning curve. Sure, really only very technical people being able to make use of these kinds of things. But we saw the biggest opportunity not in, you know, building a great platform for an IT team or a software engineer, developer to you know, make them a little bit more Productive, and building these kinds of solutions. We really saw the biggest opportunity in putting this in to the hands of an operations manager, someone who you know knows what they want to do. They know how they want to move data from A to B and how they want to change it right and what to get out of it.

Speaker 2: 43:42

So they can put it on a canvas if you give them a pen, right, but they cannot write the code or Create the API connection or something like this, and we're trying to really abstract all of that away and build it in extremely simple but still Customizable way. So, in the end, you know any task that is really taking up your time, that is repetitive, right, and that follows a combination of Rule-based logic, but also Some AI part to it where it something looks different All the time but you want to do the same thing with it. These are the kinds of tasks that we that we want to automate and give people the tools so they can automate them on their own.

Speaker 1: 44:31

You know what it makes me think of is like fraud has become so prevalent in our industry and I think about Then, you know how, like some of the common fraud things or people will have an email domain that's like one digit off right, so it looks like it's this set of an uppercase I.

Speaker 1: 44:48

It's a lowercase l for yeah right like whether it's like I've seen it on both the carrier and the shipper side, where, like somebody poses as a shipper and they're gonna like tender fake loads and which then get double bro or fraudulent Rebroker and someone takes a quick pay and none of it was real, right. No, I picture like do T Lo, the way you just explained that. I picture like that's another use case, right if it, if you can tell it like hey, my customers email domain is this. If it doesn't match this, you know this is not, it's gonna flag it right. Same thing with the carrier side. So I think and I'm glad that you, I'm glad that we kind of went through what division is and what the capabilities are, because it sounds like they're they're really there's not a whole lot of limitations. The reality is what are the common use cases and we did talk through those earlier on, but you're not really limited to that. I mean, I think this could go a really a long way, especially as Our industry gets more and more Technology based. Right, I mean you look back. I mean, ben, you and I talked about this here today.

Speaker 1: 45:59

You go back like 30 years. Nobody was operating on email the way that they are today. Right load boards didn't exist the way that they are today. It was a manual process and as we've now shifted to an email-based Organization or industry, and Even software and technology is changing to more web-based versus downloadable software, and now we've got all these cool tech and tools that are integrating in, that leaves a lot of room for for air, for fraud, and I mean I think that what what your product does and what what it can do and how it's not limited by a you know four walls around to say you can only do this I think it's that's amazing and I think you're gonna see a lot more of this kind of these kinds of tools not replacing human beings. I want to make sure that we like clarify that. But it's in it's enabling, in my opinion, in my vision, it's enabling people to go out there and do the thing they're good at, which is selling and solving customer problems, not reading emails.

Speaker 3: 47:00

That's what I wanted to point out. Right Like, this tool is very valuable for creating space, right, where you kind of should have space. Right Like it's eliminating the tedious tasks. But tonight's point if you don't use this freed up time to do the right things, it didn't do you any good in the first place. Right, it's like if I'm spending three hours a day building loads and doing this In three hours a day, prospecting, and I get three hours back and I watch youtube for those three hours, right Like I didn't really gain anything.

Speaker 3: 47:29

Right, and I'm only using this example is because it's the opportunity cost. It's fantastic, because I think this is the most, it's the largest waste of time For salespeople. Right Like, to just do data entry on tedious things Is getting in my way of, to everyone's point, like talking to more people, finding more business. Right, and I think that's really the key takeaway is that, like, efficiency is phenomenal, but you're only really getting more effective if you use the inefficiency to do more of the right things as well. And I think this is the the great first step at trying to eliminate some of these things, because these are the largest time consuming tasks that I mean. They're necessary, right, but they're not like revenue producing, right. You need them to book the revenue you already have.

Speaker 3: 48:12

I would also probably make the argument that this helps you get more business too, because most of the newer brokers I see aren't responsive enough and they're like, oh, I'm just never getting these loads and I'll look at like a client of ours. I'm like, show me when your emails are coming in. When you're responding, there's two, three hour gaps. One day I'm like, oh, and you got like 45 seconds before this loads gone and I'm like you're not even in the game when you're responding three hours later. So, yeah, not only does this save time, I think it's absolutely can help some newer brokers like win more loads and Be more effective, and it's a better appearance for your customer, right? If you're that responsive, your customer thinks you are hot shit. On top of everything, if you can pull a rate and get it back to them that quick every single time they're emailing, right, you are going to seem like the most responsive broker they've ever worked with, and to me, that's also another intangible value to using a tool like this.

Speaker 1: 49:04

For sure. Um, let me ask this too. I know we're getting towards the tail end here. What is, what is a training timeline look like? And when I say train me, how long does it take if? If, let's say, I own a brokerage and I have a bunch of emails that come in, so clearly levity's got a, we've got to set up the rules and how I want it to operate. What does that process look like as far as how it, how it learns from, I guess, the rule set that you want to give it?

Speaker 1: 49:34

Yeah so depend on the situation.

Speaker 2: 49:37

I mean, yeah, everything always depends. But, um, I would say, typically what we see with our customers is this is really new to them, right? So they might have even tried to automate certain things. We see customers that have tried to build Hundreds or thousands of rules into their outlook which just move things from here to here based on some keyword or something, and it somewhat works, but not really. Or they have tried.

Speaker 2: 50:08

You know some rpa's kind of provider before and have been burned by that, and there's a lot of. You know the pain is big, but there's also a lot of disillusionment from from other approaches. So we tend to see a little bit of skepticism from customers where they tell us I'm sure I've tried things before, they didn't work out. So the way we approach it is just tell them hey, let's take the next three to four weeks, we have a look at your historical inbox, right, what came in? We analyzed that. We run that through the system like you would in in real life, right, in real time, and we just see what the system would have done and compare it to what the human did, right, and then we can also look at how fast that the human replied, how fast could you have replied and also look at, well, if you reply within a minute, the likelihood that you get the load is this If you reply within, you know, 10 hours, it's a very low likelihood, right?

Speaker 2: 51:16

And then the way this is set up usually in the first week we just, you know, evaluate it together with the customer. Hey, how does the workflow look like? We map it out together with them on our workflow building canvas. We then run it with a bunch of examples from the past and then they evaluate, you know, do the right thing, maybe we have to adjust a few things and then usually on week four we can already go live. So we just had a larger customer, a Gulf relay, which really went live within three weeks from starting the project, and now they're automating basically all of the carrier communication by email with us. So, depending on how complex it is and you know what your requirements are, it can take a little bit longer, but usually that's how it goes.

Speaker 1: 52:12

Interesting. That's cool. Ben, you got any other questions?

Speaker 3: 52:15

No, I mean, that's awesome, though You're able to give them like really objective analysis as to where they could have been, so it really makes them have a much more informed decision. I think that's a great approach to try to getting people to understand the real value you know and return on the investments of the time and to the tool. I think that's fantastic and that's kind of covered everything I want. Is there anything else you wanted to share, thilo? Or to be able to make sure we get in?

Speaker 2: 52:38

Yeah, I think the point you mentioned about fraud is a good one, right, Because it's not one of the use cases we have done yet with our customers. But we see news left and right that fraud is becoming a bigger and bigger problem and in the end, like you mentioned, it's the same approach, right? It's a bunch of patterns that a machine learning model and some rules can be applied to look at and then flag something, right? So this could even happen just within the inbox, where you get us as an outlook plug-in or as a Gmail plug-in and just say flag it if there's a high likelihood there is something fishy going on here, right? So that's a good point. Maybe we make this kind of a free tool that people can just plug into their outlook and get a little bit of protection. So I'll definitely get back on that.

Speaker 1: 53:37

And they'll get a taste of Levity, as it is too so for sure. So levityai is the website. We've got a link in our. If you're on podcast, just check the description box or the show notes. If you're on YouTube, go down to the description. There's a link in there. What's the best way for folks to reach out to you guys through the website, or is there anywhere else we should direct people?

Speaker 2: 54:02

Sure, I mean you can go to the website, just fill out our contact form, or hit me up directly on telo at levityai it's T-H-I-L-O. Or you can also hit me up on LinkedIn. I'm usually very responsive.

Speaker 1: 54:16

We'll put a link in for your email and for your LinkedIn as well. Awesome, good stuff. Well, we appreciate having you on. We're excited to see where all this goes. And, ben, did we miss anything that you wanted to ask before we close it up?

Speaker 3: 54:35

here. Yeah, covered everything I had on my list, awesome.

Speaker 1: 54:39

T-H-I-L-O. You mentioned you went to Manifest For a lot of our listeners. They go to a lot of the different conferences. Is there anywhere that people can see you this year? That you'll be, as far as you know right now.

Speaker 2: 54:52

Yeah, I think the next one we will attend is TIA in April, so yeah, is that Technovations?

Speaker 1: 55:02

I think so Is that Capital Ideas Forum or whatever, something like that. They do a bunch throughout the year. But okay, cool, awesome, well, good stuff. Make sure you guys reach out to T-Low or check out levityai and go to the contact form and that'll get you in touch, ben. Any final thoughts here?

Speaker 3: 55:25

Whether you believe you can or believe you can't, you're right.

Speaker 1: 55:30

And until next time go bills.