AI in Freight Brokerage: Automation & Cost Savings with Thilo Huellmann of Levity | Episode 259

🎙️ Can AI revolutionize freight brokerage operations? In this episode, we chat with Thilo from Levity about AI’s potential to transform the industry. We start with a celebration of Thilo's wedding, then dive into how AI tools like Levity are streamlining communication, automating load sourcing, and managing outdated systems. We also cover the latest industry news, including Canadian Railroad strikes and LA port surges, plus insights on Fed interest rate cuts and the upcoming NFL season. Finally, we explore how automating SOPs can save costs, improve accuracy, and scale operations in brokerages of all sizes. Tune in for a deep dive into the future of AI in freight!

Support Our Sponsors:
QuikSkope - Get a Free Trial: Click Here
Levity: Click Here
Bluebook Services: Click Here
DAT Freight & Analytics - Get 10% off your first year!
DAT Power - Brokers & Carriers: Click Here
DAT Express - Brokers: Click Here
Truckers Edge - Carriers: Click Here

Recommended Products: Click Here
Freight Broker Basics Course: Click Here
Join Our Facebook Group: Click Here
Check out all of our content online: Click Here

Transcript

Speaker 1: 0:19

Welcome back. It's another episode of the Freight360 podcast. We've got a special guest with us again. We're going to reintroduce him in just a moment here, but if you are brand new, we're glad you found us at Freight360. You can check out all of our other content at Freight360.net. There's a lot of blogs, there's full length podcasts, there's some downloadable content and there's also the Freight Broker Basics course. If you're looking for a full length training option, make sure to share us with your friends and colleagues in the industry. Again, we know you've been doing that because our numbers and our audience continues to grow.

Speaker 1: 0:55

Make sure you guys are subscribed to the newsletter. You can go right to our website to sign up. That comes out every Tuesday and Thursday and leave us comments on YouTube. Email us questions. We're happy to answer all of your questions on our Final Mile series. That comes out each Tuesday, all right? Well, without further ado, we've got T-Lo from Levity back with us today. We're going to talk about some AI matters in the freight industry and the industry in general. T-lo, how are you, man Doing? Well, a newlywed Tilo.

Speaker 2: 1:26

Yeah, congrats. Thank you for having me again. Last time I was only engaged.

Speaker 1: 1:33

How's things with Levity been since we last spoke?

Speaker 2: 1:36

Good, good, I mean quite some things happened throughout the last six months and, yeah, happy to talk about. You know how things have been going on our end.

Speaker 1: 1:46

Awesome. Yeah, we're going to definitely dig into that on the show today. Ben, how are you Doing? Well, got the nice backdrop now with the TV there.

Speaker 3: 1:56

Yeah, I still got some stuff to do with the lighting, but they actually installed my blinds as we were in the green room this morning. So I think they literally I can't believe how fast they did it, Like they did the entire house in like 35 minutes, 40 minutes.

Speaker 1: 2:08

Wow, did they have everything already measured?

Speaker 3: 2:11

Yeah, they were all custom, so they basically just one guy put them all in the windows, the other guy just ran around and just fired them all up. But I was really impressed. I was like I figured this was going to be a huge hassle. So, yeah, not a bad start to the day Cool Little sports recap.

Speaker 1: 2:27

We are coming up into the quiet NFL week, the week between preseason and regular season, but looking forward to it, it was an interesting preseason all the way around. Everyone who's in a fantasy football league likely either has done their draft or will be doing it this upcoming weekend or week and we'll be kicking off the NFL season in just about a week from when this episode releases. So outside of that, anything else in the sports realm, any golf or anything fun going on that you're aware of Nothing I've seen, not yet Kilo. What's going on over in your neck of the woods in the sports world, anything fun or exciting.

Speaker 2: 3:09

I mean, the only thing that I can think of was we attended the Bitfreighter ALS charity golf event. Oh nice.

Speaker 1: 3:19

How was it.

Speaker 2: 3:20

For this month. I wasn't there myself, but two people from our team were, and they they had a lot of fun, so I can definitely recommend this one I met those guys last year.

Speaker 1: 3:32

They're like the edi integration piece, right, yeah, pretty cool, great guys, yeah, awesome, well, cool. Um, on the news round, I'll just pull a couple of highlights from our newsletter that came out on Tuesday. So the big, some of the big stories that we've been following, are the Canadian Railroad strikes. So, as of again, this comes out Friday, things can change between Friday and earlier in the week, or later in the week or not earlier in the week, but basically the the two Canadian railroads that were going on strike finally got things back on track, you know, preventing what could have been a massive disruption. Outside of that, there was a report that came out.

Speaker 1: 4:21

Really it's nothing breaking news, we've kind of seen this coming but the port surge. So the ports of LA and Long Beach are seeing their busiest times. They're saying it's near the levels that we saw during the pandemic surge. So if you remember, in the peak of COVID there was that, I think at, sometimes like what, 100 container ships just hanging out. They couldn't handle the capacity.

Speaker 1: 4:47

But the whole reason is a lot of companies are trying to like, a lot of retailers, specifically, are trying to push imports in now, ahead of what could be a potential US tariff on Chinese goods and then you couple that with a potential major dock worker strike, it could be a massive disruption to the supply chain, specifically when it comes to the ports and imported goods. So we'll keep an eye on that. Those are the two big ones I followed. Again, if you guys aren't signed up for the newsletter, please go ahead and do that. Go right to freight360.net. There's a little form you can fill out on the homepage and we send that out every Tuesday and Thursday. Anything else in the news, ben.

Speaker 3: 5:30

I mean nothing this week really. I mean I've actually been following more of it, but I mean I haven't seen much, even in the stock market or anything really lately. I mean it's kind of just been, but August is usually a slow month for quarterly earnings, for just about everything other than I think the Fed is. Well. At least there's a very high level of confidence. There's going to be a 25-point basis cut in September and I think there's also a high level of confidence they'll see three quarters cut by the end of 24, as of at least this week from what I've been reading. So I don't know. I think that's good news for the economy. I think that's good news for shipping as a whole too, and I think, as we're finally going to start the quantitative easing meaning allowing some more cash back into the system is good for everybody in the country. So I'd say pretty good news overall.

Speaker 1: 6:22

Yeah, I just have two more things. So the TIA policy forum is coming up in a few weeks in DC, so it'll be interesting to see how those discussions go on Capitol Hill with members of Congress about fraud in the industry. I'll make sure to share my feedback with you guys and if anyone um that's a follower of the show is going to be there, please let us know. I'd love to um be able to meet any of you that will be uh at the at the event. The other one too.

Speaker 1: 6:52

This was from freight waves. Earlier this week I had it in the newsletter. It was about um, so project 2025, it's the. If you're not familiar with it, it's like the, the 900 or whatever page document. Um, that was essentially you know. I think it was like a conservative organization put together as like a playbook for how they, how they would envision the US handling things moving forward, hence 2025. So this is I'll just read you know what we had in our newsletter, but it says the 900 page guide advocates for the FMCSA and the National Highway Traffic Safety Administration to codify Trump era AV standards allowing commercial trucks to operate without human drivers.

Speaker 1: 7:35

The document criticizes the Biden administration for stalling these regulations in favor of pushing Americans towards EV and public transit. So I don't know if anybody has actually read the entire document themselves. That's one of those things that I would probably throw into like a chat bot and be like hey, give me the highlights here. But whenever there's a presidential election or a changeover of administration whether it's House, Senate or all three if you include the presidency just like we talked about the potential tariffs there's always discussion about, oh, what's going to change, what's going to shake up? And the reality is things don't. They rarely ever tend to be as extreme as they're rumored to be. But you know, we'll definitely keep our eyes on how any regulation or policy change may impact our industry. So have you looked at any of that Project 2025 stuff, Ben?

Speaker 3: 8:32

I haven't heard of it until you just said that. Oh really, yeah, it's like a heavily criticized document.

Speaker 1: 8:37

Again, it's not like the you know a political. It's not like the Republicans wrote it. It's a conservative group that wrote it saying, hey, this should be. You know, if, if, uh, former president trump gets re-elected into office, this should be his playbook. But it's not like he's personally endorsed it or anything like that. So, um, yeah, but it's like it's really there's some really extreme stuff in there as far as like slashing a lot of government. Um, it's just, you know, it's very conservative doctrine, so but specifically, there is some stuff on there for transportation that was obviously worth reporting from freight waves. So we'll keep an eye on what happens. All right, shall we get into some AI talk. Let's dig in. Ben, what do you got? You're definitely more of the enthusiast on AI, so feel free and take this wherever you want to go. I know what's really cool. We've had Tilo on twice before and we've got a good success story of a client that is now using Levity. That we'll dig into in a bit here. But where do you want to start things?

Speaker 3: 10:50

off what Tilo? Where do you want to kick things off, tilo? Where do you want to kick things off? Because I'm cool with wherever I know we've got some bullet points, but I'd rather you kind of lead on what you want to initiate it and we'll go where the conversation goes.

Speaker 2: 11:02

Yeah, I think what's important is to kind of go from the theoretical talks that have been there for quite a while talks that have been there for quite a while right of you know there's so much potential. You can do this, you can do that right and talk more about actual implementations of what are people doing, what is working, what isn't. Basically, you know, cutting a little bit through the hype, through the noise, because I have quite some, you know, interactions with customers and also potential customers that tell me, you know, there's so much stuff out there and I get contacted all the time. I see so much and I don't even know where to start. I don't even know what, what actually works. So I think it's a good point, it's really helpful, to just get into the weeds of what's actually working and have some real world examples.

Speaker 3: 11:52

I think that is where we should. That's a great place to start. I was taking a look at what you had sent over even last night or earlier this morning on that LinkedIn post, and I think that post was really well written because what it does is it like it outlines the step-by-step workflows in a freight brokerage and what we and how we train a human to do it. And what I really liked about how they framed it is because this is the way I've looked at using AI is I'm like I look at it more like I don't want to say like an employee, right, but I look at it as like somebody that is an assistant that I can use. But any assistant you've ever had in any industry is fallible, makes mistakes in the same way it does, but it has advantages and has disadvantages. And if you understand what it's good at and where you know it isn't in some ways like, you can really utilize it for many things.

Speaker 3: 12:40

And I always kind of chuckle in my head when I hear somebody criticize part of it because it's like, oh well, it can't do this and I can't.

Speaker 3: 12:46

I'm like, yeah, I know, but I mean, just because it can't do everything you've ever imagined right now does not mean it doesn't have valuable use cases in the moment.

Speaker 3: 12:54

You've got to be able to see what and how you can use it and, to be honest, I think in the bigger market and you have this conversation of billions of dollars and they're looking for the ROI. This is how you get the ROI. You look at the use case and you look at what you're spending and what you could spend by using a tool that can be more efficient than training, hiring and oftentimes in our industry, turning over people over and over in this role, because some of the things I think it does very well are the things that human beings don't enjoy doing right, like the repetitive tasks, the things that are very detail-oriented or task-oriented, that are kind of honestly like from my point of view I'll speak personally like kind of mind-numbing to do for eight hours a day, you know, week after week, month after month, year after year. So why don't you want to take us through like a use case, or do you want to walk? How do you want to kind of frame it? Just to dig down into the weeds on where?

Speaker 2: 13:46

these are, let's do it. Just just following up on what you just said, right, um, one of the learnings that we also made is it's not only about the actual task at hand, uh, where you might say, okay, it takes me five minutes to do this and if you automate it, you, you know, save this much money. Right, but, like you, like you mentioned the training aspect and the high attrition rates in these roles. Um, demographic change of, you know, more older people retiring, you know, and it's difficult to fill back these roles. I was recently talking to a fairly large brokerage and they have they are still relying on an IBM AS400. Oh my gosh or AS200, something like this, this and they told me it takes six months to get someone up to speed, to actually train them on the whole system and actually do the job.

Speaker 3: 14:44

Wow, and you think about that cost just alone right there, six months, even just calling an entry-level salary right 50 grand $60,000 plus benefits. I mean like that's not a small number. And the person training them right, you're paying them to take the time from doing something else, or probably this task to do that, which is the biggest thing Nate and I talk about. In our industry, a lot is like how do you actually get more done? You still have a finite amount of hours, even for small businesses in our industry. I think this is incredibly powerful because it allows you to get more done without having to take time out of your day, which you already don't have enough of, to train somebody to be able to get more time back.

Speaker 1: 15:23

Yeah, and I'll add there too, like the you guys made really good points on, like attrition, generational changes, training and all that the tasks that we do have evolved like fairly rapidly in the last couple of years.

Speaker 1: 15:35

Because if you just look at fraud by itself, think about the things that we have to do now to protect ourselves from fraud through carrier vetting, whether it's prior to a load or at the actual transaction level. We weren't doing these on a regular basis three years ago. Right, like, if you think about and there's even some things like I don't know if you guys remember or if you guys are privy to this, but the California clean truck check program that, like the TIA gave guidance on earlier this year, saying, you know, obviously California has always been kind of a unique state as far as they have additional regulations for emissions and carbon, all that stuff. So they had this new clean truck check program that if any truck drives through California, whether it's picking up, delivering or going through it, they need to have this certification. Now that's new as of January 1st 2024. And as of January 1st 2025, they have to have an inspection for this clean truck check and they have to have this certification. The issue is TIA says. Well, as brokerages, we should be responsible for vetting these carriers if we're sending a truck through California or into California to make sure that they have this certificate, because California doesn't have a searchable database like the FMCSA to verify the certification. So that's an additional step that we have to do that.

Speaker 1: 16:49

I am fairly certain that that and other similar mundane tasks you can use a non-human being to help automate that right. The same thing goes with verifying things with, like, a carrier's email domain Does it match the registered email domain from FMCSA or is it someone trying to spoof and pretend to be a legitimate carrier when they're not to try and defraud you? So the use case goes on and on and on, based on the evolution of the tasks that we do at the load level and in general and brokerage. So I think because the LinkedIn post that you guys are referencing wasn't the guy talking about there are certain steps in our workflow as brokerages that can be replaced by the large language models. But I don't think he didn't argue that every step could be at this point right, but he was saying there's, there are use cases like I think track and trace was one of the ones that he gave um, uh, like he had an argument for.

Speaker 1: 17:47

But yeah, I definitely think that. I just wanted to add that, in that the processes that we do are different now and they're going to continue to change. Right, and having a you know, a computer learn something that won't, the computer is not going to. I shouldn't say a computer is not going to retire, but AI is not going to retire. It doesn't have a lifespan or a career lifespan that humans have. It doesn't have the turnover, a lifespan or a career lifespan that humans have. It doesn't have the turnover. So I think there is definitely a way that we're going to use this and it's going to continue to evolve over time, the same way that load boards have changed, the same way that computers have changed and all that stuff.

Speaker 3: 18:23

I'm going to add one thing and then I'm going to kick it over to Tilo to jump into the use case. I've read a lot about this in the automation related to warehousing, right, the robots and also AI being able to do fulfillment. Walmart just, I think, built some massively automated warehouse the first one that is using AI as well as robots and everyone kind of said in the media like oh, this is taking people's jobs. And then the one guy who specialized in the warehouse space said I think turnover in their industry was at 100% Meaning like literally not one person they hire lasts more than a year, or it was like 98. Like maybe one or two out of 100 last more than 12 months, because it's a job kind of you know, like you just kind of don't want, it's like it's physical, it's not like, yeah, and he's like you know, these aren't jobs where we have applicants.

Speaker 3: 19:12

The reason we need the technology is because we can't hire people to stay to do it Right. So again, I want to say that because, like we always get the criticism of, oh, like it's taking people's jobs, and the reality is like it's honestly, in my opinion, taking the tasks away from people's jobs that they don't necessarily want to do anyway and allows them to get more of the valuable work that creates value for their customers. So I want to jump into a use case, whichever one you're, you think yeah, I mean we're digging in.

Speaker 2: 19:38

First, the linkedin poster referencing is also stating, I would say, the the most obvious ones right, it was from uh santosh, um, partner at dynamo vent, who invest a lot into freight tech, and also, I think, some brokerage-type companies. Yeah, and I mean there are these obvious ones, right, like track and trace, order entry, quoting right, where there's a lot of incoming emails from shippers, where you know you need to more or less do the same thing over and over again. But there are also a lot of non-obvious ones where you have to think about, through a framework that we actually put out, is this a task that is worth automating or should be automated? And obviously it needs to be repetitive, it needs to take some manual time to be done, but then also it needs to be something that you can verbalize. If you're able to put it in an SOP, then that's a good sign that it can be automated. And then it needs to be digital, in a way where you can pull the data from the inbox, for example, and you can push it to the TMS. So if all of these things are given, then you can do something.

Speaker 2: 20:56

And actually one of our customers that you actually referred to me. He wanted to do something very obvious, which was just new orders coming in in his Gmail inbox and putting them into Thai. He's currently adopting Thai, so we were looking at that and then, while we were talking about how this all works, he was like, oh, actually, can this also be applied to this use case and it was a very non-obvious thing where he has his team on extracting all of the BOLs from his inbox and then checking who the shipper is, because it might be, you know, that they want to Prospect them.

Speaker 1: 21:41

Yeah, prospect them.

Speaker 3: 21:42

They might not be working with everyone, right, and that's the thing that like, because, like, I've talked with him about this over the years and it's like one of the lowest hanging fruits for a freight brokerage is being able to prospect one step from your customers.

Speaker 3: 21:54

Right, you've been working with this customer for years, but we get so busy in the day to day of making sure freights picked up and the details and just the job that, like it is very easy to overlook that like, oh, that was a new delivery, that could be a new prospect. We should reach out because it's an easy call to make hey, we were delivering there. How did that go? We work with a mutual customer. It helps initiate the conversation. There's already trust there. But even at these large companies that's the one thing I've worked on most when I work with managers or brokers that, like the CHs, echos and TQLs is like just people overlook it and it's like if you just take the time but oftentimes you don't have the time or it's not organized and it takes forever to go back through and to cross-reference and I think it's a really good use case.

Speaker 2: 22:37

Yeah, and he was already doing that with his team manually right, where they were going through the historical emails, they were looking at those PDFs, they were basically, you know, copy pasting all of the info about these strippers and then putting that into Rocketreach for outreach and then into HubSpot as a CRM. And we're bringing all of this together now right by just automatically scanning the inbox for these types of documents If it's such a document, extracting exactly that info, enriching it through Rocketreach and then putting it into the CRM through Rocketreach and then putting it into the CRM. So the whole workflow is now very condensed, where they can already start working with that very, I would say, valuable data set being pre-prepared for Earth and save all that time on the manual copy pasting.

Speaker 1: 23:25

I'm curious did you guys figure out what the estimated time per transaction that that will save?

Speaker 3: 23:33

You've got to imagine it's got to be like minute, multiple minutes of pol right, I know the labor and I don't want to like disclose all of the details, but it's significant, like because they've been doing this for a while and they have done well with it, but it is incredibly time consuming and takes a lot of individuals a lot of time to do this, for sure yeah, yeah, and there's.

Speaker 2: 23:53

there are these non-obvious use cases like this one, right? So I think you need to look at it through the lens of what are my people doing all day long, right? Where are they spending most of their time? And then, in these brackets, right, is there anything that is very repetitive and could be automated? Right? And oftentimes there is, yeah, yeah. So here's one that.

Speaker 3: 24:19

Go ahead Nate.

Speaker 1: 24:20

No, I was going to say I'm curious what your guys' thoughts are, because we talk about processes that you should automate and maybe some that you shouldn't the reports of the companies that have. It's like over 100,000 AI sales calls were made so far from a brokerage. And I'm curious, I'm still not sold on that one yet. I, like for prospecting, I just feel like it's not yeah, I don't know that the technology is there yet. I don't think that it is.

Speaker 1: 24:49

Yeah, yeah, I'm curious what you guys think on that part of it, because I mean there definitely is certain phone calls that can be automated, like a check call, right, or a customer update call or something. You know the same way you get a call it's like hey, reminder, you've got a dentist appointment on wednesday at 10 am.

Speaker 2: 25:02

you know, like it's, you know you don't need to have a back and forth conversation, but the sales calls is where I was like I don't, I don't know how I would feel about that one yeah, I think the one that you're referencing um is uh, it was actually not sales calls and uh, I do also think that, oh, it wasn't sales calls, no, it was check calls, updates, things like that Inbound calls.

Speaker 3: 25:24

They're inbound calls. Yeah, I've demoed it with them.

Speaker 1: 25:28

Yeah, in that case. Yeah, that's cool.

Speaker 3: 25:30

But it's close because I've talked to another company. There is some products out there that are doing to Tilo's point structured outbound calls that can ask a series of questions to grab some information before they send it to a human being. I've seen some of that and I have seen it handle inbound calls to filter carrier calls in to then distribute them. The use case and I think it's a good segue because I want to ask Tilo this the use case I've been thinking of recently and I'm curious if you've thought about it is there are so many emails related to covering loads right, and, yes, there's some automated email systems, even on a tie, that we use. Or I can send an email to every carrier in my system about a load that's in there, if they're in my system. But I've got access to much better sourcing tools right where I've got large email lists, and to send an email out to three or 400 carriers on a lane. If I do that five times in a day one, I've got an issue on outflow.

Speaker 3: 26:26

But secondly, being able to interpret what's coming back becomes so cumbersome that it's incredibly difficult to work through and I've always thought that would be a really good use case for levity, because to be able to manage, get the outbound emails out and then to be able to sort through and ask basic questions like when would your truck be available? Can you confirm your equipment type? Just being able to filter those few things to get the number from 50 emails coming out to the 10 that can actually pick up your load, I think would be a phenomenal use case, because that is incredibly cumbersome for people to do 100%.

Speaker 2: 27:01

Incredibly cumbersome for people to do A hundred percent. And actually this was one of the first things that we implemented with one of our customers. They wanted to get started with quoting first on the shipper communication side, but then they decided to take a, take a turn and just do the carrier communications. And it's exactly that use case there. They have their network or their email lists where they know, okay, they cover these lanes roughly right, and then they blast out emails.

Speaker 2: 27:30

Replies come back right. Either they say, yes, we can take it for the price that you're looking for, or no, we don't have any capacity right now. Or, yes, we can take it, but for a different price, right? And in those cases, uh, what? What we do then is uh, first of all understand what kind of reply is it right, um, and then extract the relevant information, like the rate, for example, um, and then put that back into the, into the tms um, yeah, and that way they they basically just say hey, we want this load covered. They click one button, emails go out, they wait half an hour, they get stuff back and then it auto populates their TMS. They can just choose who they want to do the. Yeah, do business with.

Speaker 3: 28:16

I want I want us both to make a note that I we're going to connect after the show. Us both to make a note that we're going to connect after the show, because I don't want to talk about it on the show, but I'm going to connect you with some folks that I think are going to have one of the best sourcing tools out there and I think you guys working with them could be a game changer from the user side and I think that's going to be a really good connection. So I just made a note, literally that after we'll connect about that?

Speaker 1: 28:39

Can we real quick? So I know we've talked through in previous discussions some of the tools that Levity plugs in with Can you give a recap on? Obviously we've mentioned email and TMS platforms and even pricing. Are there any new specific vendors that you guys have connectivity with or is there any new types of platforms that Levity have connectivity with? Or is there any new um types of platforms that levity is working with?

Speaker 2: 29:07

um, yeah, I mean one one that is very common that we see is as green screens, uh, as you know, domestic truckload rating engine uh, very common and uh, they have a great api. So it makes it very easy for us when we see someone new coming in saying, yeah, we're using Gmail and green screens and tie right. This would be like a great combination for us because we know it's very, very easy to get started with these tools. Bitfreighter, I mentioned earlier right, they cover the whole EDI part, but then you know, there's always some customers that just can't or don't want to integrate via EDI and we basically bridge the gap to their EDI platform by saying, ok, we pulled whole stuff from the emails.

Speaker 3: 29:54

I want to pause there. So anybody else out there because Nate and I we've gotten this question a lot right, like for smaller brokerages that are growing right and you're starting to land some larger customers. This is a big sticking point right For growing companies because, to your point, not all of them have the EDI capability or the tech team to support it. But you land a customer right, or your customers, you're negotiating the agreements and they're like well, hey, our requirement is EDI, right. I think that's a great reason for anyone out there that is working with a customer that requires EDI, that might not have that capability, to reach out to you guys at Levity because, just to your point, you can bridge that gap until maybe they do, or either way, like I think, to me it's absolutely as effective and it's probably I don't want to say easier, but like, I'll let you kind of walk through the difference between having to do an EDI and being able to use Levity for that.

Speaker 2: 30:48

Yeah, yeah, I mean we're not an EDI provider, right, or we don't help with that particular piece, but BitTrader does right and they are a great partner of ours with, I would say they are a great partner of ours with, I would say, quite good pricing, also for smaller businesses, because they do it on a per-shipper basis, right.

Speaker 2: 31:11

You pay a certain amount, you have a shipper and then, no matter how many loads they're sending, it's always the same price and it's pretty affordable to get started with. So I can definitely recommend checking out bit trader for for edi stuff. Um and uh, yeah, I mean we then cover the parts where you know it's still email based, right, and that that happens, right. Sometimes it's even like emails being sent out and then you get a link to a bidding portal off the shipper and you have to go there and then fill out stuff and you know it just. It's not something that can easily be done if it's high volume and you have a small team. So that would be the part that we cover, but we then pipe it back into Bitfreighter, which then you know.

Speaker 3: 31:55

How Walk us through that use case? I was doing that yesterday, so like, how does that actually work? You mean the bidding portal stuff or yeah.

Speaker 2: 32:00

So like how does that actually work? You mean the bidding portal stuff, or.

Speaker 3: 32:03

Yeah. So like, let's just say right now, like I mean, I literally did this yesterday, right, we had a bidding portal get sent over. There were 400 lanes on it that I manually had to go in, log in, I pulled up 400 lanes, I exported that into Excel, I then had to manually reformat it into DATs format, put it in their multi-lane and then I had to work through based on because this is for next month, so like I can't use today's rates, I had to use rate cast to next month and then I've got to factor in the fuel, the mileage, next month's estimated rates, pull all of those back and then I've got to put them back in the old format and then I upload it. Like that was literally two hours of my day yesterday, oh wow. So how? Or I mean walk through a user how this could look from their point of view, I guess?

Speaker 2: 32:50

So how did you get the initial request? Was it an email?

Speaker 3: 32:55

Email came through with a link in the middle. Press the button.

Speaker 2: 32:58

Ah, okay, okay, yeah, I mean the way we would approach this is, first of all, you always have this unstructured piece of data. It could be an email, but it could also be, you know, a document, a web page, whatever.

Speaker 3: 33:14

A file whatever.

Speaker 2: 33:15

Yeah, what we do is we identify okay, it's a list of loads, right, we get that list. It's a list of loads right, we get that list.

Speaker 3: 33:23

We identify the different attributes for each load, Basically pickup, dropout, all the variables, equipment type, pickup delivery, maybe commodity.

Speaker 2: 33:35

Then, for each hit, some sort of API Could be. That could be green screens. Bitfreighter is actually building something on top of these to not only get a rate but also decide. Do you actually want to quote that? Do you want to have certain rules applied to it depending on your internal workings, right? Rather than just getting a rate so for every single one and then close the loop by having an automation that goes on that page and just fills out what's needed on the bidding portal so I don't know which one it was it needs to be something that we can support. It's not always the case, but if we can support it, then you can automate that step as well. Right? So it's the pulling, it's the understanding, it's the enrichment from a rating engine and then pushing that back to bidding portal, for example.

Speaker 3: 34:33

So question if I wanted to. Let's just say I have, and to your point, criteria for a brokerage. Let's say, most of my customers, most of my carrier base is in the Southeast, for example, right. So I want to focus on lanes in the Southeast. Now let's say the bid's national. So there's 500 lanes in there across the country. I know my biggest advantage is in the Southeast.

Speaker 3: 34:57

What Nate and I teach people to do manually is to go into your existing TMS. Look for where you have good rates and good carrier relationships. So the Southeast, for this example, right. The rest of the country maybe we want to bid on them to just put our name in the hat, but we know we're not going to have a strategic advantage. But we want to participate.

Speaker 3: 35:12

So in theory, tell me if I'm understanding this correctly, right, like we could automate this. So basically, the first step would be to take this large bid. It goes and grabs all the lanes, looks at all the criteria we could then have it, identify all of the lanes that we actually have lane history in our TMS, quote them from our lane history on what we're paying plus whatever margin we want, and then we could say okay, anything that you don't find a hit on. We want you to now go to that and just take 9% or 10% above median rate and it would get me the best rates out of my system on what I'm actually paying and who I have carriers for. And then we could take the second step to fill out the rest of the bid to at least participate in case there's a fallout from someone else that we still have an opportunity to get that load. That's a feasible use case.

Speaker 2: 35:57

Yeah, yeah, it does require some setup, but it's definitely doable. And it's also similar to what we've seen where it's like every single brokerage has different, you know, is differently specialized as a different carrier network, right. And the last last week I was just talking to one that has basically their own system to to to decide, based on a zip code, whether they actually want to, you know, quote something or not, right? So it's like, if it's a one or two, then yes, we definitely want to take it. If it's a three, maybe if certain you know other uh criteria add up as well. If it's a four, we don't want it at all, right? So, yeah, um, that's what they basically created. And then now every email that comes in extracts the zip code, checks it against that list, right, and then either quotes it or not, can I ask you one more step?

Speaker 3: 36:51

If it can do that, then my guess is, since it can email right, could you layer on one more step and have it email all of the carriers that ran that lane and say, hey, this is a bid we're participating in. Would you want to give us a rate on these volumes and to have it literally now go and reach out to the carriers that I've already had lane history with to ask them what they would want related to the volume in the bid?

Speaker 2: 37:14

Yeah, I mean, let's keep it simple, right? If it's three-digit zip to three-digit zip, right, and you have your TMS lane history and you would just want to pull the top five that have had the most loads in the past I don't know 90 days or so Then you can pull that list, you have their contact information. You can put that all together and send out the email right away and close the loop that way as well. So you both have the coverage part and the coding piece.

Speaker 1: 37:44

I think, yeah, I think, ben, the way that I'm grasping it is like we said earlier, if you can create an SOP for the process, you should be able to automate it Right, and what you just described was step by step. It was like, if this, then this right, and it's basically like a flow of the process and that all makes sense. So I'm curious, though, tilo, what happens when there's an exception or like, let's say, a zip code comes in and it's either not a correct zip code or it's a misformatted Canadian postal code, or it's missing a digit? How does levity handle exceptions?

Speaker 2: 38:27

I mean there are exceptions all the time, right, and this is also an understanding that I think some people are missing is it's not zero or one? And when you, when you're automating stuff, it's more like automate the easy pieces where it's very obvious and you, you know there are no, you know questions, right. And then you have some hard cases where you know someone needs to look at it, uh, evaluate things, maybe, call someone or whatever, right, like there are exceptions all the time, right, but you need to put the uh, the rules place so that you know, ok, it's not following that happy path anymore, right? So someone needs to look at this stuff.

Speaker 1: 39:06

OK, yeah, that makes me you're right, because it's the same way that, like if, if 100 percent of the stuff I like how you said, zero and one meaning like automation versus human, is human right.

Speaker 1: 39:26

So if your current process is 100% human, then you're going to normally have stuff that has to get escalated to someone who can make a decision about it. And the same thing sounds like it would be applied using an AI tool like Levity, where, hey, if it can follow this process simply and easily, you can automate it. If it's an exception or requires escalation or doesn't fit inside of the rule set that we've identified and put in place, a human being is going to then take action, but, depending on how you have it set up, you could automate 50% of a process or 75% of a process or 25%, whatever you decide you want to do. However you implement it, it sounds like it's a good way to augment the human workflow or work like your pool of workers, so that you can refocus that time and energy on other tasks that you don't necessarily want to automate, like sales.

Speaker 3: 41:35

Prospecting being able to make more calls. It's the biggest bottleneck and growth for every brokerage is they get bogged down with tasks and they can't get to the phone to be able to reach out and make new relationships. Right. Every time you're freeing up another 30 minutes, right, like that's another few calls you can make. It doesn't seem huge, but again, that's compounding growth. Do that every day right Over a year, like you're getting a lot more work done.

Speaker 3: 41:58

That's where the lack of productivity or effectiveness or efficiency is in most brokerages and to me, that's the thing most people don't ask, or companies don't ask is they always know what they want more of, but very rarely do they ask what can they eliminate to create the space for the more? And it's always like well, I can't get rid of anything, I just got to do more and I don't have any more hours in the day and I'm like well, okay, like let's go back a step. What can we do to create the space for the more? Right and to me, like that is how you get the ROI from a tool like this is being able to do those things. Can I ask you, what would you say is the most common use case? You?

Speaker 2: 42:45

I think the I mean when, when we talk to these companies, right, they either already have something in mind or they are just generally interested because they know they're wasting a lot of time on certain tasks. But I think the one that gets prioritized in many cases is is the quoting piece, because it has multiple effects, right, has multiple effects. Right. It does not only have the effect of you saving time and putting that into prospecting, but it also has the effect on you know, win rates, right, if you are able to answer in one minute instead of 20, then this probably has an effect right. So I think it's often prioritized. But you need to have the pieces in place, like a very clear way to rate, to find a price, right. If it's all over the place and you have five spreadsheets, then we can't help you. Right, you need to get the basics done first. But if it's in place and you're already using green screens or you have your own thing right, then it's pretty straightforward and it has the I would say, the biggest leverage there.

Speaker 3: 43:47

I want to say there's two things that I think are relevant. One, this was back probably like 2019 or 20, like right around when Nate and I first started the show. I was working with a brokerage and this is when GPT came out, before chat, and we built similar where it was able to take stuff out of the email, put it into like an internal load board. That was API to that Right, and what we were able to do then which is what you guys are doing now was we had one sales guy basically onboard every lumber company in the entire country, all of the jump off rate, the stuff that you're talking about, which is you get the rate in first because you have a truck. Then you win that load and it's not huge profit margins Like maybe you're making 150 a load but if you can do that at scale, you can really do well, and we were able to have basically that automated, which is what Levity does now grabbing this information as the emails come in immediately, getting them posted to DAT, refreshing them, and then just have two people answering the phone.

Speaker 3: 44:42

As carriers called in that were a fit for that load, we were able to book like tens of thousands of dollars a week in GP and a side of the business that everybody moves away from but they start at because it's just too time consuming and, to your point, you're not fast enough to get a rate back or a truck to win it most of the times. So it's, you know, it's an opportunity cost. You go like well, hey, you know I don't make that much for the time I'm going to move to a different customer, but there is money if you can scale and find the efficiency right. To me that is one of the best use cases for levity.

Speaker 3: 45:12

These loads where, like you say, you get a quote, well, you're on with a customer, you're doing something else, you might have the right truck, it might be at the right time, but if you're on the phone and you miss that and it's 25 minutes later or 40 minutes later or two hours later, someone else got that load and you could have been the better fit for that customer and been able to capitalize on it. How does that kind of work from the user's point of view? Like they just reach out to you and you guys kind of like set this up, Like how does that work if someone's interested in doing this?

Speaker 2: 45:46

Yeah, we do help people a lot when it comes to setting up their first automation, because it is something new, right? It's not like an established software category, like a TMS, where everyone knows what it is and how it works. So it's new to people and we wouldn't want to just throw them into the product and, you know, figure it out. I've had it. So we do have an onboarding team that you know sits down with you and oftentimes it actually comes from the SOP. So they say, hey, hey, we have 10 people, they are following this process, here's the sop. And then we sit down with them and translate that into an automation.

Speaker 2: 46:16

We run that against historical cases that have already been handled by humans and just see, you know, is it doing the same thing? Are there any discrepancies? Oftentimes the discrepancies that we find the ai was actually doing the right thing and the human was doing the wrong thing, because a lot of people they don't have the time to correctly input all of the data. Sometimes they just click, click, click and be done with it, right. So we set that up and then when the first automation is running, then you know they're off to the races. Usually, right, they understand the concept, they start building their own stuff and then, over time, automate more and more small things, right? Even if one thing just saves you I don't know a couple thousand bucks each month, right? If you add that up, you come to something that's quite valuable.

Speaker 3: 47:05

I want to point out the other side of this too, because, like, when we are I mean, nate and I have done this like training a human being to do this, right, like everyone goes, oh well, like I can just train somebody and I'm like, but the thing I don't think most companies take into account is to get somebody to even proficient not anywhere near the accuracy you're saying right, Like it takes three, four months. Okay, I mean, someone else is spending their time. There's a cost that person is getting four or five grand a month at least for three, four months. You're 12, 15, $20,000 in training somebody just to be able to get them to do this, where they can reach out to somebody like Levity, reach out to you guys and you can have this up and running. And what's the time frame from, you would say, cradle to operational?

Speaker 2: 47:50

I mean, we've seen customers go from zero to their first automation implemented within two weeks, if it's a straightforward case and they have everything, all the tools in place. Sometimes it takes longer due to you know they have their own system, you know they need some API build right and then there might be some delays there. But yes, it's fairly easy to set up and it's also one of the things that people are most surprised about, because they're like oh yeah, this is super complicated and you know you need to think about so many things, it's so technical and it actually isn't anymore right. Maybe it was.

Speaker 1: 48:31

Ben, there's also I mean you definitely hit on, like, the financial costs with training. There's another level to that, though. So like, yes, if you hire a human being, so obviously first there's the cost to pay them.

Speaker 3: 48:45

The interview before that too, which I just reminded. When you said that, like you got to talk to 25 people before you find one and that takes time, I wasn't even going that route, but you're right.

Speaker 1: 48:52

So there, okay, there's the. There's the recruitment side the cost to acquire them, a cost to train them and pay them when they're not actually producing revenue for you. Then you like you kind of outlined like the opportunity cost for the trainer, who then doesn't have the time to focus on their revenue producing tasks. And the part that you didn't mention, that I want to hit, because I've seen this with people that hire and they poorly train or they don't fully train or whatever the cost of mistakes that an employee makes. Right, like we had a guy that he ended up getting let go. Right, like we had a. We had a guy that he ended up getting let go. Basically, an agent had hired him and had him working, had this guy working on some of the basic dispatch operational tasks and he started. He would screw up, like maybe once a week or twice a week, but we're talking like six hundred dollar mistakes every single time. So not only are you paying him now, you're also losing money and it's like. It's like it's like a double hit.

Speaker 1: 49:50

So, um, you know, I think AI has come a long way and, again, like the, the use case, the generic use case we mentioned, is like if there's a standardized process and there's a flow to it, follow the steps right and you you're not going to have human error. So I just wanted to add that in there, because don't forget about the cost of losses, right, and you can really blow that up into, like if it leads to a claim because a rep hired a carrier and didn't do the right thing and sent the wrong equipment in, and next thing you know it's our fault and now we're out $50,000. Thing and sent the wrong equipment in and next thing you know it's our fault and now we're out $50,000. Like. I have seen this happen.

Speaker 3: 50:31

I've seen a client of mine had this in the past two months. We ran into that hired two experienced people because they were trying to, you know, mitigate the training costs. So huge salary, right Made the mistakes cost just literally money. Blowing out the door for two months until they could finally realize what you were saying, tilo, like you could do in a test model, like being able to run it backwards and see like, okay, like this is what could have happened. This is what did happen. Is that what you wanted to happen? No, we didn't. Okay, like it's also highlighting from a training point of view, where you can help train the people that will stay there to improve and automate the piece so that you've got less mistakes, higher accuracy, better efficiency, better effectiveness across the board. I think.

Speaker 2: 51:17

Yeah, 100% yeah. There's a lot of opportunity out there.

Speaker 3: 51:23

You know, the other thing I wanted to talk to and I was curious, both your thoughts on this. We talked about this on the show over the past year or so, where CH had dropped a lot of their salespeople and Nate and I kept going like how is this possible that a company that large is cutting their sales staff this much? And I do not know if this is for sure. True, but I've heard from a number of people over the past few months that they were building an in-house tool very similar to what levity does to absolutely be able to not have as many people staff their customers.

Speaker 3: 51:54

And ch I think for anyone out there that isn't familiar with how ch works, is like they're a low-cost freight brokerage that doesn't have great service, like I think that's always been the reputation in the market.

Speaker 3: 52:03

They're going to be cheaper, but you know they have a higher rejection rate, usually on their freight because, again, like as the as the market moves, they can't just take loads for large losses. That's been their niche. So to me like a product like that fits really well. That's how they're able to literally save hundreds of thousands, if not millions, of dollars in salaries every year and I'm like holy shit. I was talking to Natos. I'm like this is literally like the largest company in our industry is using this. This product is available from you guys at Levity to any size brokerage out there right now can harness the exact same tools. The same companies with the biggest budgets that can utilize it, like and access it or need to understand it, because there's people like you that are going to be able to help hold their hand and show them how to implement these things and walk them through this Right.

Speaker 1: 52:58

That's a really good point, ben, because I mean we've obviously over the years we've talked with all different sorts of vendors and solutions providers and we've all heard of some of the big ones out there and the reality is some of the technology that's out there, some of the TMS platforms that are out there, you just can't afford if you're a small brokerage. You're just not there. It's designed to do way more than you'll ever need and with that comes a cost and that may be a flaw on the company's business model to not be able to service the smaller, midsize company. But clearly the the client use case that we outlined before he's not like a hundred million dollar brokerage. You know what I mean Like. So, yeah, I think that's a really good point, tilo, unless you had anything to add in to that part I want.

Speaker 1: 53:45

I wanted to talk about some future stuff and kind of where you see AI going, specifically in the freight domain, the freight space, like the next five years, the next 10 years. Obviously, we've lightly hit on in the current time. Here's what we can automate, what you should maybe should not automate things like that. Do you have any, you know, in your crystal ball? Do you have any idea of where, whether it's levity or you know, not related to levity where do you think AI will have use case or biggest impacts with freight brokerage, specifically in the like distant future?

Speaker 2: 54:26

Oh yeah, that's that. Yeah, if I knew, I think I would probably, you know, I don't know You'd be designing it right now, right? No, but like it's hard to tell, right, you need to, I think, constantly react to you know new stuff that is going getting out, right, just seeing, okay, how good is voice AI? Now? Right, how good is this? Right? Can I actually apply it? Or is it still too early and not working yet, right? I think it's difficult to say. But I mean, I saw Brad Jacobs talk about kind of like the future of of freight in general and he was mentioning like at some point, right, uh capacity will be traded like the stock market there's like you know, you buy futures and and you do all that stuff.

Speaker 2: 55:21

it's just like the next logical evolution of a more efficient and forecastable market in a sense, where you know, uh, certain manufacturers or you know produce uh people, they just they, they want to kind of maybe have an insurance right on on having certain prices for for shipments in the future, things like that Interesting.

Speaker 1: 55:49

I had a thought last week and this is not very related to the examples that you just gave, but I just thought about it. I wanted to mention it. I've seen a lot of times where companies get frustrated about how their employees may send an email out. So there's a lot of animosity between brokers and carriers sometimes and you get like into these email feuds and I always almost thought, like what if an AI tool made insured for a company owner that every one of his or her employees or representatives sent professional emails and like would cause? You know how some people write emails and they don't mean it but they come across as like really rude.

Speaker 3: 56:30

Just the tone at anything like.

Speaker 1: 56:33

Oh imagine if that email was crafted in a way that comes, gets your point across through a filter professional way, Right, Maybe it's going to be called levity email. You know levity email tone.

Speaker 2: 56:45

Yeah, it's interesting, I think. I think we built something like this once in a hackathon at our company where it was like, you know, everybody is communicating differently. Some people are, you know, very direct and it might seem rude to other people, whereas others are communicating very nicely, right, and just having that, you know, having like an invisible translator filter or something in between that just makes sure it's coming across in the best way possible to the receiver, but transporting the same information, something like that, yeah, not sure if it's something we will build anytime soon. I just thought about it.

Speaker 1: 57:24

You know it's funny. You said that.

Speaker 3: 57:27

Nate, I had somebody that was having this issue, a client that was having this issue with one of their brokers, and what we were having them do was, hey, before you send the email, copy it in to GPT and ask it what's the tone, what the tone is Right, and literally ask it to give it and then copy it back in before you send it, because it was happening constantly and I'm like, well, this is at least, I think, a good training tool because they'll start to see what they weren't seeing in their own verbiage. Right To everyone's point, this is so hard to teach a human being. I read tons of books on communication. I'm reading one on marriage too right now, because it's something we can all keep working on and just how we communicate.

Speaker 3: 58:04

And there are things I learned literally yesterday and tones in the way I phrase things and I'm like, oh, I didn't realize we were having an argument and I didn't realize what I said actually initiated, and I'm like that was certainly my fault in the way I phrased it, didn't realize it. But it's very hard for us to understand where these habits have come from in our entire lives and to ever catch them. And even if you can catch them? How do you practice it enough to change them, moving forward to have a better impact, whether you're in sales or talking to your best friend or your family or whoever, and to me that business use case is, I think, really interesting yeah.

Speaker 1: 58:40

Well, good discussion. We'll definitely have to get you on again in the future. Tilo. Any final thoughts here, Any closing points you want to make before we wrap it up?

Speaker 2: 58:53

No, I mean, I'm looking forward to a road trip that my co-founder and I will do next month. So we're starting in Chicago meeting with a bunch of brokerages and freight tech people, and then we just drive down through the Midwest Tennessee and also stop by in Texas and yeah, if anyone wants to meet.

Speaker 1: 59:15

let me know. We'll put your contact details in the description box or in the show notes, but it's levityai is the website as well. Awesome. Well, Tilo, thanks again for being on with us. Ben, any final thoughts Whether?

Speaker 3: 59:29

you believe you can or believe you can't you're right, and until next time, go Bills.