Welcome back to Unboxing Logistics. I'm your host, Lori Boyer. Glad to be here from EasyPost. And today, I have an exciting and interesting guest for us. We have Jonathan Salama. And he is joining us. And I, we're going to be talking today all things TMS, brokerages, all of these cool things that maybe you know about, maybe you don't.

But I have brought John here today because he is going to teach us everything we don't know. John, you'll learn quickly with me, there's more I don't know than I do know. So, happy to have you here. John, introduce yourself to our community.

Thank you for, for having me. I'm Jonathan Salama. I'm the CEO and co-founder here at Transfix.

I don't have your traditional logistic background. I come from the world of tech and data science. I'm originally from France. Moved to the US for college with the intention to go back home. I met my wonderful wife a year after college and was never coming back. I started Transfix about 11 years ago now because I met my co-founder who had this idea for at the time, what we thought it was going to be very easy and was, was no, no roadblocks whatsoever.

I can tell you weren't from the logistics world.

Nope. Just thought that that has to be a fairly easy, manage matching shippers with trucking companies, what can go wrong? But it's been a wonderful journey, and I'm happy to be here.

Awesome. Okay. So, before we dive into our topic. I want to know a little bit. This season I've been asking my guests, what is your go to comfort food? You know, the phrase comfort food, the food you want when you are just, mmm, need something that'll make you happy.

That's an easy answer, it's pizza. Is there, is there another answer? It has to be the only answer.

Pizza is my answer, John. Now, I've heard all kinds of answers. John, I don't want to tell you this, but somebody, whom I love, said salad salad is

My wife. My wife would say,

Oh, see, that's who I wish I was. I wish I was the person who went for salad and comfort food, but I would be, I'd be splitting the pizza with you. Yeah.

Yeah, happiness comes from pizza. I'm pretty sure.

Absolutely. So, okay. We are going to be talking, as I mentioned, you know, brokerages, TMS, just data technology, how all of those, obviously tech, you said you come from this tech background, how that is changing the industry. I actually really love, isn't it cool, everyone, how John came in from outside the industry.

I think that those kinds of perspectives really bring so much insight and just help us look at problems from a different angle. So I love that. At the beginning of every episode, I like to make sure that if somebody had to drop off, if somebody left, that they come away knowing, you know, at least a couple of takeaways.

What are the most important things that you would want someone to learn or remember from today's session if they didn't remember anything else?

Leveraging data to drive not only just decisions, but repetitive actions that can be either automated or drive for specific outcome. I think that's what we specialize in and what we see over and over again is this notion that I don't have enough data.

Or it's too messy. There, there's, there's always ways to, to understand and, and amplify it. And I think using third party tools or leveraging others, other knowledge is, is really critical in this day and age to get to what we need to do.

I love that. I am a total data nerd. I adore data, but data can be very overwhelming for people, and data can be really-- my husband is the type where he gets data overload really easily, and he doesn't know where to start. So I'm going to be excited to kind of pick your mind about. I think most people want to leverage data. They don't always know how and the best way.

So we're going to get into that. Anything else that you would want people to come away with today? Or is that the key factor?

I, I think the key factor: leveraging data and technology.

Okay, perfect. So I want to start talking first, but a little bit about brokerages. So tell us about your journey, how that kind of shifted with brokerage, you know, what transfixes today. So there are just for a little background for our audience, so they can understand what perspective you're coming from.

And then, a lot of our audience may not exactly understand the role of brokerages. So I want to kind of cover that before we dive into everything else. So can you share a little bit about that for us?

What is a brokerage? A, a brokerage is it it is a company. The sole purpose is to match shippers and specifically their shipment with trucking companies that could execute their shipments.

It is, I wanna say almost like a marketplace, and I say almost because most, if not all, marketplace, are supply driven? Well, this is very much demand driven. If you have a shipment, you will usually find a carrier for it. The real question is that what price will you find that carriers? Brokers make their typically makes make their margin based on the difference on which they sell the truck to the shipper and which price they buy the truck from the carrier.

A difference is how how a broker makes, makes their margin. With that in mind, our thesis 11 years ago, when I met Drew is that brokers, it was true 11 years ago, and it's still kind of true now, are highly inefficient. They're very good at what they do, which is finding carriers and talking to them and negotiating with them.

But, there is a lot of paperwork that needs to happen in between. There is a lot of step that needs to happen between finding the shippers, Bill bidding, bidding on the freight, winning and creating the shipments in your system, all the way down to delivering and invoicing. There is tons of step that is happening that where a broker creates their value.

And why I believe shippers should use brokers that there is true value in using brokers. So, so our thesis was. That we were going to automate and use data science to figure out, like, how do we make every single step either fully automated or the human that is doing these steps, the, the person behind those phone calls is highly, highly efficient and has every tool at their disposal.

So if the average at any typical brokers is 12 shipments per employee, we wanted to be at 50 or 100 per employee. That was, that was our thesis. Over time that developed more into a cost per shipment and we found out that the the, the average broker spent around like 150 to 200 dollars per shipment in man hours, while we brought that price down between 25 and 50 dollars.

That was, that was our, our whole thesis from the beginning and what we've developed through the years.

So John, sometimes brokerages kind of get a bad rap, you know, people don't always think the best of them, but you were saying, for sure, you've obviously, and obviously you built a brokerage. What are the benefits?

When should people are, you know, I guess. It seems like they'd be valuable for companies, but there may be some companies where brokerages don't make sense. Can you kind of walk our audience through pros and cons, you know, so that they can know about that.

In my opinion, every company should be using brokers.

It is highly, highly time consuming to execute these transactions, to make sure you find the right carriers, to pay the right price for the carriers. There is value for brokers at any, at any steps. Now, to, to be fair and transparent, I don't have any horse I forgot the, the expression, I think, is I don't have any horse in the race.

Yes.

And sometimes American idioms are tough.

No, that was perfect.

The, the, We are no longer a broker. We sold our brokerage to NFI and became a tech company supporting brokerage. Which is why I can say honestly, I actually think brokers are helpful to every shippers out there and to every carriers out there.

It's very true, some broker has given brokerage a bad rep. But I think that's true for every profession or every group, individuals that there is like, there's always some good apples and some bad apples. For the large majority of brokers that we've talked to, or we're interacted with, they generally want to do what's best for shippers and carriers. And I think they provide a very, very important service to both.

Okay. Are there any signals if people are looking for a brokerage? Are there any red flags, any signs that they could look out for to say, like, maybe this brokerage wouldn't be the right type of brokerage or, you know, something that would help them?

Yeah, I think there are a lot of tools out there that helps you understand what is the proper market rate. What is the proper rate for a specific truck? And, going to a broker that either is way below or way above that rate is definitely a red flag and often Way too often, a way below the market rate is considered like, Ooh, great brokers. I'm going to go with that broker. And this is where, this is where trouble happens. There's a reason there's a market rate. There's a reason you should stick to it and not pay significantly below.

Yes, I think that is so right on. It's the lesson we all know in life. When something seems too good to be true, it probably is so be careful with it.

Okay, so it sounded like what you're saying, from brokerages, one of the main advantages, at least early on, were those kind of relationships and their ability to negotiate those kind of things. But you're saying that you think that it's more reliant or we need to have a focus on data today. Like, I guess, how have those kind of merged?

It's so if you're a shipper, think about like every time you need to find it, especially if you're a small shippers and you have maybe one shipment out of one facility on a weekly basis, finding a trucking company that's going to execute that shipment is like taking a dart on a on a map of the U.S., hitting it, and you have to hit that exact point at the exact time that you will find a truck there.

Yes,

Gotta be really hard versus a broker who can aggregate that volume, use data to understand, All right. I have this many trucking companies that I have relationship with, that have likely a truck available, at this area, and this set of shipments from shippers and again aggregate those shipments. And, and make these relationship on a macro level versus a one time.

And this is where data comes in. This is where understanding the market rates comes in. This is where brokers can be highly more efficient at finding these carriers than, than shippers trying to, to hit the dart on the board at the right time.

Have most brokerages moved to this kind of data model, or are there some who are still slower?

They're certainly trying. There, there's, it is really hard. We spend a whole decade with a team of data scientists that every day, every day that I work with them, I'm more impressed by them. They teach me every day.

I want to be a data scientist. They sound so cool

They're, they're really cool. But I hope they're not all getting like massive ego boost from this.

Stop listening. Stop listening.

But yeah, it is our new business is trying to enable these companies who wants to be understanding their own data and their own buying power in the market, and helping them make that, that make that step.

It's really hard. Most of them have a from talking to a lot of them, they have a hard time understanding where to get started, how to understand their vast unstructured data. And that's why working with a partner is going to be key for them.

How has this kind of digitization, you know, the, the data coming into brokerages, how has that impacted kind of logistics and supply chain in general?

I think overall it's made it move a lot faster. And it's, it's, it's, and it triggered, or it is really helpful in, in an era post COVID where a lot of people are working remotes. A lot of people are working very, very differently than what we use the way we used to work just two or three years ago. Having ability to understand a market much faster has created a lot of leverage for everybody in the space that we didn't have before, whether it's from the brokerage side, the shipper side or the carrier side. And I think today's speed is what you demand from everything you use in your day to day life. If you load Netflix tomorrow and it takes 10 second to load a whole movie, you're going to be frustrated when in hindsight, you're loading a whole movie on the computer in under 10 seconds.

That's really impressive. Especially if you're doing on a flight, I tend to be very frustrated with flight, with flight WiFi when that's really impressive tech. So this in the same, same thing happened in our day to day as a, as a logistic expert, you want that accessibility. And it's finally at at your fingertip.

It hasn't been it. It's definitely evolved in the last decade, right?

So I want to talk about integration, then, with kind of other technology. We're talking about tech, and so specifically with your transportation management systems, your TMS how do they kind of compliment? Do they work in tandem? How do brokerages and TMSs kind of work together?

Yeah, usually, they typically integrate with each other really well. So we have for transport specifically, we have a TMS for shippers, and we have a lot of tools for brokers. The the TMS for shippers have allows for a better aggregation of data so you can understand the performance of each broker that you're using. The beauty of the brokers, you don't have to use just one. You can compare them with each other and understand their rates, understand their performance, and make sure that you're using the right one for the right loads. A TMS can help do that. A TMS can help organize All all the communications that are happening on on any given on any given shipments, which typically is a lot.

There is a lot of moving part on just a single shipment. It is every time I walk into a store, I'm impressed at how all the stuff got here. It took an army to get to, to get all those goods into the consumer's hands. And it's impressive.

Totally agree. The first time I went into even a warehouse, my mind was blown as I then thought of it on scale. I mean, it's, it's incredible what we do in this industry. So speaking of TMS, I've heard a lot of talk about maybe like the unbundling of its TMS. Can you explain kind of first what that even means and why people are talking about it?

Yeah, so, so I think what, like, the first person I've heard say this is Dave Broehring from NFI. I'll give him credit for that term. Yeah. That's a good word in his mouth. But I think what I believe that means is no TMS today is good at everything. There, there's no one TMS that it gets, especially for broker, less so for shippers. There's a lot of them out there, as I mentioned, we have our own. They're a lot more established.

The ones for brokers are not. They're usually the afterthoughts. They're good at maybe one thing, but very bad at everything else. Or they're mediocre at everything. While the unbundling effect is just the ability that we have today to take pieces of the life cycle of a shipment, and take one stage, and create one tool that is very good at that stage, and apply it to a multitude of TMSs, especially all the proprietary TMS that people have connect to those. And make that specific part of the of the journey, either highly automated or highly data driven or highly efficient in one way or another.

I'm just, you know, I'm gonna play devil's advocate here. So it seems like that might be challenging, though, if we unbundle everything, you know, does our tech stack become really giant? How does it work with integrations? You know, what are the challenges that come with that?

Yeah. There are there are tons of challenges. That's why that we don't, the, the, the whole structure hasn't happening just yet. It will probably happen in the, in some future, but it allows you to create, or it allows us to create a TMS that over time can have feature that can be turned on and off and that can, that can be very central to everything. That can be good at everything and can be very flexible.

Customizable maybe.

That's right. To the need of a specific broker. Rather than trying to be rather than creating one platform that should fit all the needs of every broker, start creating a platform that adapts to the need of the, of the broker.

I love that. I recently had a conversation with somebody who is just a really forward thinker in the industry. Always. I'm always so impressed with him that he is always ahead of the game and kind of it's like, I don't know. He knows what's coming. But one of the things he says is with technology, it's so critical that we can kind of have in place what we need now, but also the infrastructure or the ability to then adapt and add and take away As our needs grow. And that's how we think forward into the future without paying too much, maybe now.

So that kind of sounds maybe along the lines of what you were saying?

Yeah, yeah, yeah, I, I think that aligns directly with what I was saying. I think that, that, that infrastructure at the bottom or, or the, the building blocks is it's critical to get right. And I think that's what we aim to do.

John, you're talking to us a lot about how you've used data science and machine learning and all these cool phrases, you know, as you've switched from brokerage over to your, to your TMS .

Well, what does that really mean? What does that look like? I, I think sometimes we hear words and they just sound really cool, but they also seem like, I don't even know what that means. So, help, help the people like me.

So, that's a great question. I would We've been doing, we've actually been doing data science for, for almost a decade before it was cool. Before it was cool

I was just going to say, John's been doing AI since before AI was cool.

The, there are two aspects of AI, and there's actually many, but for the purpose of this conversation, there's Gen AI, which everybody's talking about today. That's a cool thing.

That's chat GPT type things, right?

And then there is the machine learning models that we're building to understand what rates may look like in the future. Talking about the first one first. There are many application to it, and you're seeing it in your day to day life without even noticing. Some emails you're getting, some response you're getting that are generated by bots that are highly efficient at creating sentences. And, and some brokers today are deciding to put this in front of, to like, read their customer email and have responses. So the response time is, is faster. I personally have a controversial take on this, that you should not do this.

It is 40 percent of the time wrong at best. I, the, the wrong conversation with the, can lose a customer. But you should be, you should be spending time. In developing relationship, not automating them.

Okay, I'm going to say it one more time. We shouldn't be, we should be spending our time developing relationships, not automating relationships. Okay, sign that off as, you know, John Salama, you got it. Yeah, okay.

Love it. And so I, our, all of the Gen AI that we've developed serves internal purpose, purposes to summarize, all the communication that we've had on their shipments to create one quick paragraph so that our team doesn't need to go through 100 emails to understand what happened and can just understand, okay, this is the summary of what happened. I can talk to my customer quickly while the customer is on the phone. What happened on the shipment? I don't need to know. I look at the summary, and if the summary looks really strange, I just refresh the summary so that I'm not saying something that doesn't make sense.

Having that filter to create that relationship, but make it efficient is where we think Gen AI, that's how we've been using it. That's a very small portion of how we use AI. The biggest portion is the way I mentioned is around our machine learning models. I, there is a few third parties out there that will create an index or a market rates for any given shipments or any given lane. A lane being a city to another city. Understanding that rate is critical, but understanding as a broker, your buying power against that rate is a lot more important and understanding your buying poet power in six months from now, when seasonality either at a peak or a trough or you know that in this region there's going to be it's hurricane season or so forth, having that understanding ahead of time helps you set up 12 month RFPs, or it helps you set up a proper spot str.. Strategy for a specific shipper.

Understanding what your buying power is, is really, really hard to do when you're asking a human, because if you're asking two people at the same company, working different region, what's your buying power on the other regions, they're likely will have a different answer.

Yep. Yep.

And so having consistency into how you work with any shippers or any carriers, create trust. And it creates predictability in your business, which is Really really important when you're building a brokerage. Predictability is probably the hardest things we're all we, well, for us, we were, but brokers are dealing with.

So, okay, I'm going to throw a couple of questions just at you. I, again, I'm a data person. I'm sure we're the same, John, because I love AI. I love, I just can geek out on the imagination of the future and the potential applications of it. But the truth is a lot of it's not quite used yet in the industry. You know, we have a lot of hype about it. But sometimes we hear it's overused or... What, if, if I were going to ask you, where do you feel like AI and machine learning really like today make the biggest difference? In, in what kind of aspects in the industry?

For brokers specifically in predicting either spot rate or contracted rate. And understanding your own costs. I think that is absolutely critical. And that's what everybody's using. The, having the one thing that we've been using a lot that is less number specific and more directed to carriers. We offer carriers recommend what we've been calling Recommended Loads using using some of the recommendation engine that you use in your day to day life.

So if you go shopping and you put this white shirt on in your cart, and you see, ooh, you may also like these blue pants. That algorithm like is a very well-defined algorithm in e-commerce, but was really hard to tailor to logistics and, and we're starting to move towards towards a world where your, the day-to-day interactions you have with AI that you don't know is AI is appearing in logistic more and more.

And so we we're using these recommendations. We're using, as mentioned, all those machine learning models for range recommendations. But there's actually there's even I was talking to a company the other day. They are doing training using AI. So the, the. If you, you call in as a sales rep and the bot on the phone on the other side is pretending to be a potential customers, and you have to close them.

There's another aspect where I found super cool.

That is way cool. Okay. I totally agree with you. I always say rates, so whether you're a brokerage, whether you're just a shipper and you're compare, you know, AI is... you are not going to get ahead unless you're using AI to compare. That is just such a cool way.

I hadn't thought about the recommended load. So, give an example. So, maybe if you're shipping a certain lane or something, that the algorithm could predict this other one. Explain that to me.

So, if, if we see a carrier that does a lot of New Jersey to Chicago loads or or or or some loads like this. We could start understanding, well, like, maybe they'll maybe they'll like a load coming and picking up in Philly. It's not the same region. It's not the same market, but it is on the route. Maybe so. The algorithm is a lot more complex than this. I'm just trying to like give a concrete example.

No, that was that was so perfect. That was a great example.

And in a way I hadn't thought about with ecommerce. We see that all the time and how that kind of segues over. So do you feel like there's any differences? Let's say you're a small business, fairly small, versus large. Are there differences in maybe the ways you should be using AI and machine learning?

I think if you're, if you're really large, you can, you can set up your own algorithm. You can you, you, you should be leveraging your own data over others' data. Third party data is a lot more. Well, if you're, if you're a smaller shippers. Using third party indexes and getting market recommendation from third parties who have been aggregator for a very long time are probably the way to go.

But at the end, whether it's an aggregation of others data or your own data, you should be acting the same way against it. And so that as you grow, you'll have that ability to ingest and to understand, to focus on data.

Yeah, absolutely. It's all about putting that into place so that you're ready for that growth in the future. Completely agree with that. I feel like everyone listening, If you are one of those kind of data resistant or just start trying every day, because like John was saying, you're interacting with it all the time, and you don't even know. So just start playing with it, get used to it, and, and you'll you'll get more and more comfortable with it and start to see the amazing possibilities.

But we're kind of on the heels of RFP season, and I'm curious about how maybe you know, data science plays into that when we're, we're going through that process.

Yeah, I think it's two-fold on this one. It's it's definitely data science, but it's just sometimes there's a there's just we do, we group the management of data and data science into the same bucket. While, I think data.

Okay, say it one more time, John.

Data management and data science are often put in the same bucket. As the same thing. But data management is just the ability of just using the right tool to understand what's going on support for a shipper. Like RFP management just means comparing all the rates that you're getting from a carrier and being able to manage them properly so that you can pick the right one. Cheaper one the cheapest one might not always be the best one. But very often, as people, that's what we gravitate towards.

So having, so once you have the management of data, getting data science on top, guides you towards which one is the right one is what's important after that. Right, so I think shippers in particular, like what we've seen when we were a broker, we're doing a lot of manual work in Excel and not utilizing enough tools to just manage the process. And if you don't have that, it's really hard to put data science on top. It doesn't, it doesn't work.

I, this could have been one of your takeaways for me at the beginning. But having the data and managing the data need to be kind of seen and dealt with as two separate, but equally necessary things that you should be doing. So you need to have the, and sometimes you have the data that you don't even know, right? There's a lot of systems out there. You need to have processes in place for both managing and making sure you have accurate data.

John, you and I, I'm sure are, nothing irritates us more than inaccurate data. So having the right data and, okay, that is so, so smart. Anything else you want to say around RFPs?

No, that, that, I mean, in a nutshell, I think that's it using the right tool.

We're almost out of time, but I want to make sure we talk a little bit about the future. So what do you see as kind of the emerging technologies? What are the trends that, you know, looking into your crystal ball, John, what are you seeing in the future here?

There there's a lot of cool things that are going to happen in our space. Whether it's Gen AI getting better every day and eventually going to be helpful in our day to day lives. But I am a lot more excited about automated trucks. Automated truck, like the, all those AV trucks. I think I think that is going to revolutionize our space.

I don't know how long it'll take. I'm no expert in the field, but that's I'm particularly excited to see where that is going.

John. I wish it would come a little sooner because I have teenagers learning to drive, and I really wish they could just be in automated cars, because my life flashes before my eyes. Maybe by the time you're you're you're girls are that old.

I hope so. There's a whole decade before that happened. So we'll see.

Yeah, exactly. Come on, everyone out there who is working in the automated vehicle. We're relying on you.

Yeah, we need it.

Okay. Thank you so much. This was really, really fascinating. I love the idea of using data science. I love the idea of making sure you're managing your data, taking a look at it. Don't let it overwhelm you. Again, my community always knows I say, you know, pick one or two things that you can do and that you can implement. Don't get so overwhelmed that you do nothing. You know, don't get paralysis from it.

So, John, where can people learn more about you, learn more about Transfix and what you do? How could they contact you? Are you on LinkedIn?

Yeah, I'm, I'm on LinkedIn. I'm easy to find but we also have just our, our website is transfix.io. And we have a form right there. We were very active on it. So if you want to know more, just reach out.

That is perfect. So again, thank you so much. Any final thoughts or goodbyes?

No, thank you for having me. This was, this was a lot of fun and always happy to talk data.

Awesome. Okay. We'll see you all next time.