Lori Boyer 00:00
Welcome to a special edition of Unboxing Logistics. I’m your host, Lori Boyer from EasyPost, and today we are gonna be talking about a big event that just happened last week, Parcel Forum. Really exciting. We were the AI in Logistics sponsor of the Parcel Forum Show. I had the absolute pleasure and privilege of speaking with the one and only the Unboxing Logistics godfather Tom Butt, and we were able to talk about AI.
So we’re really excited today because we’re just gonna give you a few little things, what we heard at Parcel Forum, what all the scuttlebutt is, and kind of just some of the highlights of the things we talked about at AI. Tom, before we jump in, reintroduce yourself. I know that you are a, a classic favorite here at the Unboxing Logistics family, but for those of who happen to have missed one of your appearances, tell us a little bit about yourself.
Tom Butt 01:09
Yeah, sure. So I appreciate the, the privilege of coming back. I always love talking on the podcast here. So a a little bit about myself. Tom Bott, a senior manager here at EasyPost of AI and analytics. I’ve been in the operations and technology space going on, gosh, 15 plus years now and have worked with major retail organizations to come up with strategies to leverage technology to better execute operations specifically for supply chain, direct to consumer.
And with EasyPost here in, in the parcel space. So have the privilege of diving in headfirst to the AI topic, which seems to be all the buzz these days.
Lori Boyer 01:52
Yeah, absolutely. And I mean, let’s just start there. What was the buzz you heard on the floor of Parcel Forum? What was everybody talking about?
Doesn’t have to be AI. Just, you know, what, what, what were you hearing from everybody in the industry?
Tom Butt 02:08
Yeah. Some, some key themes in just talking, you know random one-offs at the conference were around uncertainty in the, the supply chain space, the ever-growing complexity the ever-growing expectations of consumers in terms of how they get their packages the different methods that they want them.
Everyone seems to want everything right now. They want fast. They want it cheap. And so the parcel space in particular a lot of the shippers are getting squeezed to, to do more with less. And it really bodes well for AI. And everyone in the AI space was talking and kind of comparing notes and saying this AI thing, is it a buzzword?
How are you guys using it? When is the starting gun gonna go off? And so we, we talked about that during our session.
Lori Boyer 03:01
Absolutely I that, those are similar things that I heard and conversations I had. Everyone’s kind of in the same place where there’s a lot going on right now and as we’re just about kicking off peak season here for 2025, uncertainty, I love that you said that because that’s kind of in the name of the game is everyone’s not exactly sure what’s gonna happen and absolutely right. I, I felt like as well, there are a lot of people who are kind of like, yeah, I know AI, like I should be using it and it’s awesome, but I don’t really know how, and I don’t really, you know, is there’s a lot of confusion around where and how and, and to implement AI.
Did, did you hear that as well?
Tom Butt 03:47
For for sure. You know, there are a lot of tools out there. It seems like every single platform you go into is like, oh, try this new AI feature. Whether or not that’s something like a Google Gemini or if that’s, you know, just ChatGPT or could be anything from, you know, summarizing calls and taking notes.
So I, I think a lot of it is there’s just this amorphous like, hey, AI can help us, but how? And, and so what we were really talking about during the, the session that we had was, you know, start small, start very targeted and say, how, how can I use AI to help me with a very specific problem? You know, outline your objectives, outline expectations of what you’re looking to do.
And by doing that you can be a little bit more targeted, a little bit more strategic about how you use it.
Lori Boyer 04:40
Yeah, I loved that. I loved hearing people talk about AI is really effective when you start with your problem. So just what you said, like don’t just try to squeeze AI into a pigeonhole just because you wanna use AI, but if you have a problem, which, I don’t know if we ever don’t have a problem in our industry.
Tom Butt 05:01
Right. That’s one of the things I, I love about the operation space is you can always get faster. You can always get cheaper. You can always provide a better customer experience and that that frictionless buying experience, if you’re going direct to consumer or if you’re doing B2B, you know, you’re your customers, you’re looking to just make things just sharper crisper, more clear, those sorts of things.
Lori Boyer 05:22
Yeah. I also, somebody mentioned being able to see your ROI. I think that that’s another critical one when you are looking for AI tools, you should be able to measure if you are saving time or saving money things like that so that, you know, it’s not just a buzzword, not just hype.
Tom Butt 05:41
Yeah. And you know, with that one of the things we talked about was, you know, creating a scorecard, looking at it, breaking down by di different dimensions speed, cost, quality, what you’re doing to support sales.
And what you’re doing to help support the customer experience you know, come at it with a kind of a, a framework that you can assess and say, hey, for each of those dimensions, how far am I off from where I would expect to be or want to be? And then that’s gonna help you prioritize what sorts of things to go after first.
So again, just set up a, a framework, come up with an intention and, and go from there.
Lori Boyer 06:22
Yeah, absolutely. One of the areas that I felt like the, you know, our, our session attendees were really interested in was when you talked about kind of the specific areas or pain points. So Tom here at EasyPost deals with helping our customers with AI and so he actually kind of sees on the front line.
A lot of what, where they are using AI, how does it work, how does it not work? Where do they see hiccups? So the audience was really interested in hearing what are the kind of pain points that you’re seeing where AI is being effective, where you are being able to use that scorecard and kind of track things.
Can you share that with our, our Unboxing Logistics family as well?
Tom Butt 07:03
Yeah, for sure. So again, with the uncertainty in supply chain all the factors to consider you know, we are seeing just this continued push for carrier diversification. You know, carriers are offering different service levels.
There’s a lot of different moving parts and pieces. And so with that, we’re seeing customers use AI. We have a specific tool that helps people sort through all of the, I’ll call it granular or very particular instances, so of, of how to get things to customers faster, how to get things to customers more cheaply.
And in particular kind of getting that intersection of both faster and cheaper. So we’re using AI to just crunch through tons of different data points, billions and billions of shipments. And what that’s allowing customers to do is, again, get, get the best of both worlds. In the past, it was either cost or time in transit that you’d be looking to optimize for.
But we’re seeing our customers use AI to, to really get the best of both worlds. All of the different factors to consider, you know, different zip codes to deliver to surcharges dynamic landscape of, of carrier rates. AI is really helping in that space. There’s also maybe a little bit more I call it sentiment analysis.
So people looking at summarizing customer experience, going through loading their, their customer feedback to, to suss out key themes of, of what might be going on. That’s gonna allow you to have a, a better pulse on where your opportunities might lie. So there’s that hard, like, hey, we’re gonna go after these specific metrics and targets, but there’s also this distillation of sentiment that you can get to that is gonna also give you kind of a roadmap for how you would, how you’d navigate the parcel space, especially as we get into the peak season.
Lori Boyer 09:05
So, okay. So what I was hearing was really good uses of AI when it comes to kind of carrier selection, route selection, service level selection to make sure that you can optimize for both cost and speed. And so when you said consumer sentiment, so kind of what if, how people are feeling about, how your customers are feeling about the experience that they’re getting. Correct?
Tom Butt 09:31
Yeah, exactly. And you know, if you’re loading customer comments or if you’re using AI to do you know customer experience or customer contact, customer support calls, you can use that, that AI to kind of layer in and analyze that, to give you summaries and things of, of that nature. I know in our sales opportunities, we’re using a product called Gong that allows us to make sure that we’re capturing exactly what the customer is looking to get at, allows us to reflect on how do we best serve that customer. So that’s something that can, can be extended across AI tools.
Lori Boyer 10:11
Absolutely. That was something as well, people, we had questions about which tools do we recommend?
Feel free to reach out. I have some lists of tools that a lot of companies are using. Of course, here at EasyPost, we love our Luma tool for checking your that kind of time in transit, spend, figuring out carriers. Carriers are our, what we live and breathe every day. So that’s where our expertise is. But there are a lot of tools that are really great out there across the spectrum.
Tom, do you have any specific stories or examples you can share maybe of customers or people you’ve talked to on, you know, their experience going through implementing AI, what that looked like and, and how they were able to, you know, create their scorecard or, or do anything like that?
Tom Butt 10:56
You know, we’re, we’re getting a lot of customers that are looking at how can we use AI to make better predictions about what carriers maybe that I’m not using today, but carriers that I, I should be using.
So we have customers where they’re maybe just using or, a couple of carriers, so UPS, FedEx but they’re really, you know, Amazon is a new player in this space in terms of expansion from just their, their in-house networks. So we’ve used AI to ask you know, basically our, our customer data to say, hey, are there if we brought Amazon in or if we brought another carrier and maybe a regional carrier, like Better Trucks or, or OnTrac what would it look like and what types of shipments would we use or what types of shipments would best be leveraged on those different carriers maybe that we don’t have yet today?
So it’s a little bit of a, a scenario builder that allows people to kick the tires on how they can expand or how they can augment their carrier network the impact that that might have. And it’s, it’s really looking at every single individual package that you had shipped in, you know, like a la a last 30 day period and give you the insight to say, hey, this is the value that it could provide if you bring it in.
So we’re augmenting you know, maybe a, an analysis that you did in the past we’re using lots of different data points. We’re using the AI to make recommendations on how you could augment your network.
Lori Boyer 12:29
Do you see a specific, like, kind of general percentage of, you know, how much they can reduce in parcel spend?
I know that obviously numbers are gonna be totally different from customer to customer, but are we talking 10%, are we talking 1%? Are we talking 6942%?
Tom Butt 12:49
Sure. I give you a, a generic number wide range. We’re seeing anything from a, a couple percentage points. So you know, but for somebody that’s shipping a lot of volume, that few percentage might be tens of thousands, hundreds of thousands of dollars a month. But we do see some customers, depending on, you know, maybe they’re only using a single carrier right now and they’d want to expand to a multi-carrier service, they could save up to 25%.
Lori Boyer 13:17
Yeah.
Tom Butt 13:18
So it, it is quite a range. It kind of depends on where you’re starting from.
If you’re already highly optimized or using AI to some extent already there might not be as, as much juice for the squeeze. If you are coming from a place where you haven’t really touched your network in a while, and there’s lots of opportunity to expand, to make adjustments you know, we are seeing it in that 20, 25% range.
Lori Boyer 13:41
Yeah, I, I had a great conversation at Parcel Forum with a woman where she has a, a flower selling business and she’s she ships about 10 to 15,000 packages a month. But she’s always used a single carrier. Right. And for me it’s just like, like so much opportunity. If you are listening and you happen to be somebody who’s really stuck with one or two carriers, that is a big signal that you probably have a good opportunity to really optimize your carrier mix.
Now Tom, I get a lot of questions, so, and, and this was a question she had for me. I’m scared that if I move away from a single carrier that I’m not gonna be able to get good rates ’cause I put all my volume together so that I can negotiate rates. How, how do you respond to, to that question?
Tom Butt 14:37
Sure. I think we can still look to, to use carriers in a lot of the instances to make sure that you’re getting volume discounts.
That said, there may be a, a subset of your shipments where we can start small. We can look at how much volume would be shifted for things like two day or next day shipments. So we, we can kinda adjust the amount of, of volume that, that you may be looking to shift or to expand to multi-carrier.
It doesn’t have to be, hey, all of a sudden you add another carrier, it’s 50% has to go this way. The other 50% has to go the other way. So you, you maybe lose those volume discounts. So you can start small and you can start optimizing subsets of your packages. Maybe you’re getting the most customer complaints on those expedited services.
You could optimize with AI that just those expedited services. Save the bulk of your shipments to go through your, your main carrier. But I, I’ll just reiterate, if you are in single carrier, even if you’re getting those discounts you are at the whim of that carrier and any sort of surcharges or price changes that they have. You know, there are instances where, you know, if it’s a unionized environment, maybe the union goes on strike and then you’re scrambling to, to go other places. So is there is that risk mitigation factor. But you don’t have to go, you know, dive head first. You can start to dip your toes in the water and, you know, maintain your, your volume discounts and, and go from there.
Lori Boyer 16:11
I love that and I love the idea of starting with somewhere you have a problem. So with AI in general, this is what I’m hearing. And you know, as I’ve spoken to different experts, start with a problem. Again, always start with the problem. So if you have, exactly like Tom was saying, maybe a certain area where you’re getting delivery windows are missed by 15% or more.
In a certain region or a certain lane. First of all, you can use AI tools to kind of determine where you are having problem spots, but even if all you do is know that you have a problem spots, you’re probably not gonna get much worse. So start by experimenting with some different tools or, or options, maybe different carriers here.
And you know, when you have a lot of data around it, that makes it so it’s a lot easier for you to see, you know. If you’re able to improve even 10% or something like that. I love starting with that small data rich problem. That to me is the easiest way to kind of segue into making a little bit of a change.
Another area though, Tom, I hear is like, maybe it’s repetitive or slow. Maybe you’ve got a lot of manual stuff going on. Those are other opportunities for AI. Is that what you’ve heard or seen as well?
Tom Butt 17:28
Yeah, I think those, those decision points you know, where you’re making, you’re having to adjust one-offs here, one-offs there, and it’s consuming a lot of your time.
I like to frame things in the, you know, the Pareto’s sense of, of it. So maybe you only have 5% of your packages that are late. But those may be taking up, you know, 90, 95% of your time to to break from the 80. 20 rule, to go to the 95, 95, 5 rule, or the 90, 90, 10 rule. So a, a lot of times when you are having those, those instances you can pop AI and look at your shipments, see what some of the exceptions may be. See if there’s, you know, opportunities for delays within your carrier network. Where are those occurring? Is it, you know, pre-transit exceptions? So you’re printing a label, but it’s not scanning into the carrier network.
You can take a look at those types of shipments. You can have AI analyze and say, hey, are those happening for a certain carrier set? Or are those happening kind of across the board? You may have different different solutions that you need to, to go after. But AI is gonna help you identify where those exceptions are and potentially make some recommendations on how to address those things.
I know that we talk specifically about, you know, our, our AI solution. But you can, you know, pop things into ChatGPT as well and, and ask more strategic questions there. So there, there are some things where you can start to at least, you know generate a framework of how you wanna approach a pro a problem and then use it as you kind of go through the discovery journey.
I know that there’s, you know, buzzwords past, I know data science was a big one a couple years ago. Obviously analytics, you’ve got cloud. Those are all things that we see as, as iterative. So you still need a lot of those things. You need, you know, cloud to be able to aggregate all your data so that AI can analyze it.
But I, I think in general again, you can use AI kind of throughout that discovery process and you can come up with the scientific method to you know, solve your problem. You know, again, start with that problem statement, but then kinda step through that journey with AI as, as a partner.
Lori Boyer 19:55
Absolutely. I wanted just to jump into mistakes that we commonly see and I’m gonna kick us off. And then Tom, you can share any other mistakes you see. But one of the big mistakes when it comes to AI is underestimating the importance of your data being good data and clean data. Do you have any recommendations?
You’re just a data guy in general. How do you make sure that the data you’re using or you know, that you have a single source of truth even because a lot of times different teams are using different data. What is the importance of data? Why is that important, and how can you kind of try to make sure that you’re getting clean data?
Tom Butt 20:38
Sure. And I’ll, I’ll use the play out, played out mantra garbage in, garbage out. I think, you know, most of our listeners have probably heard of that. We have seen AI be very effective at anomaly detection. So, for example there may be true outliers in your data, but if you’re pumping a data set into AI and asking, hey, what are my, some of my statistical outliers you can also use AI to simulate supplemental data. So if you only have you know, a lo a light amount of data you can use AI to generate additional data sets that, that you can reference. But using it for anomaly detection is something to at least get you down the journey of seeing how clean is your data.
If you can just highlight, you know, things that may be, you know, a couple standard deviations outside of an expected value. I know for us, we’ve got some cartonization software that’s out there, but, you know, you can ask, hey, what are, does, does this data look clean? Like, is it something that needs, you know, is it something that we should adjust or we need to augment to be a little bit more robust?
But you can use some AI tools out there to, to assess that and to just ask, hey, where, you know, are there opportunities for me to clean this data up?
Lori Boyer 21:56
Yeah. So use AI to, to kind of make sure you’ve got the clean data as well. I love that. One of the great things about AI is the fact that it sort of takes these really technical, difficult things and makes it easy for just regular people like me to understand it.
So there are tools. Again, if you need specific names and tools, I’m not gonna jump into them all here ’cause I’ll forget some of my favorites. But there are tools that help you look at your data and, and give you kind of the probability that maybe you’ve got some red flags in there. What other mistakes do you see though, Tom, when people are trying to implement AI?
Tom Butt 22:29
To go back to not having an intention, not having a, a process. I, I think that.
Lori Boyer 22:36
You see a sparkly and you’re like, ooh, that sounds good.
Tom Butt 22:41
That’s a big one. I, I think the other one is just not getting started, right? So to, to say, hey, I, I don’t know this. We’ll kind of wait for others to get this figured out.
I can tell you right now, there’s not gonna be a starting gun. And those that adopt this adopt AI and start to build it into some methodologies, build it into frameworks are gonna have a head start and a leg up. Those are the, the customers that are gonna say, hey, I, or excuse me, the, the users that are gonna say, hey, what happened? Why are we so far behind our competitors? Why are we losing market share? Why are we not as competitive from a a cost perspective? The other thing that we do see a lot of people say, well, I don’t know exactly what the AI is doing, so I’m not gonna use it. Right? So, so for example you know, you see that the AI is getting you better results.
It’s very good at, you know, you have an objective statement. It will optimize to drive toward that objective. So that could be, you know, cheaper shipping, faster shipping in the, in the parcel space. But I think if you don’t, if you don’t have a process to start to incorporate AI and you don’t have a trust that the results are gonna be out there, those results are gonna be formatted to just say, hey, this is what I’m achieving. A lot of times people wanna see the exact how and sometimes AI isn’t as great at doing that.
Lori Boyer 24:15
Especially when you get into the data teams. You get the ops teams correct. That are like, well, let me see this. ‘Cause we’re used to, I say we, like ops teams. I, I’m your, I’m your buddy. I’m not really ops teams, but yeah, you, you need to know that people feel that need to like have control, I guess.
Tom Butt 24:32
There is a little bit of a just take, let, let the AI make some decisions. Again, sometimes it’s not gonna be able to explain that the reason it chose this particular shipment method, you may have to go back and say, well, there was a holiday here, or we’re not seeing this statistical performance in aggregate from one carrier in one particular lane.
So we do see a lot of people wanting to be very prescriptive and to basically have the AI do what they were doing in the way that they expect to do it. And this is a non-linear jump, so it’s not gonna be like, hey, it’s gonna do things exactly the same as, as I did it. You know, maybe I had a really complex rule set that I, I built out.
It may just make decisions on factors that there’s so many things that it can crunch. It’s not gonna give you the exact secret sauce. But it’ll be a good tasting sauce.
Lori Boyer 25:29
Even calculators are like that. You know, if you’re adding 324 plus 192, it just gives you a number. It doesn’t say, oh, I carried this one, and I did.
You know? And so at some point we had to kind of trust the technology that it’s working correctly. I, that’s a really, that’s a really hard one for some of us, so, love that idea. Anything else there on mistakes? Otherwise, I wanna hear kind of just your takeaways on what people should do.
I, I love the idea as well, I don’t wanna miss this that people sometimes just don’t do anything, like analysis paralysis or whatever, trust that, look, when you start using AI, there’s probably gonna be some hiccups. There’s gonna be times that it didn’t work the way you thought, or there’s gonna be times that you didn’t realize or you didn’t understand how to use it.
That should not keep people from starting and implementing. ‘Cause as you said, the, it’s already started. The, you know, you said the gun’s not gonna go, the gun’s gone off. You know, people are running to, and, and the only way for us to learn and to figure out where it is gonna be implemented is to actually kind of dive in here.
So I love that point. Tom, any other mistakes? If not, what are kind of takeaways of, you know. What, how can they do, what can they do to get started this quarter? Any, any like, go out and do this?
Tom Butt 26:54
Yeah. I, I think part of it, and we talked about this during our session, was leadership. A lot of people are using AI in their day-to-day.
But a lot of times leadership just doesn’t really know, again, to, to go back to the starting gun hasn’t started or hasn’t gone off. I think part of it is as a leader, you’re expected to have a vision for some of this stuff. So, so some of it is just you know, making sure that there’s an owner of this and calling out that there’s a responsibility from a leadership team perspective to, to go after and develop an AI strategy.
A strategy doesn’t have to be perfect, but if you have some sort of framework, some sort of vision for how your team can use this, can use the AI tools. You’re gonna just be able to iterate on that. You’re gonna be able to, to build on it, learn from your mistakes, kind of go through that process of, of discovery.
So I think again, just not having a leadership vision is, is a big mistake.
Lori Boyer 27:59
Yeah. There, well, in our session we talked about, there was a, a study that was done recently. And it is leaders. So if you’re leaders, you’re getting called out a little bit here. It is leadership that tends to pull back, where team members tend to want to use AI a lot more, and leaders get a little bit more hesitant.
So it’s a good reminder actually, right before this podcast, I had a meeting with my team. I’m a leader of a team as well, and we were discussing AI tools and what we need to be using and how we can, and so make sure that you’re, you’re bringing that to your team as a leader. That’s a, a great point, Tom.
Everyone, it was great talking about AI. Some big takeaways for me. Just from today, but also from Parcel Forum, and from the session, is that we just gotta get started. AI is not perfect yet. Okay? We haven’t found like the silver bullet AI that does everything, but if you are not doing something, you’re getting left behind.
There are a lot of great tools out there. But start with your problems. So if you’re having a problem in the warehouse, look for a tool that’s gonna do something in the warehouse if you’re having carrier issues or if you’re wanting to really save some money on your, your carrier spend as we come through peak, we’ve got a lot of big surcharges and stuff.
You know, look at tools where you can simulate that and, and figure that out. But just start. And then figure out from there. That’s my advice. That’s really what I’ve been hearing is people are starting and seeing some really good wins. Any other takeaways from you, Tom?
Tom Butt 29:34
No, don’t, don’t let perfect get in the way of better. You know, it’s.
Lori Boyer 29:40
Love that automism, don’t let perfect get in the way of better.
Tom Butt 29:44
Yeah. And, and again, just build on successes. If you can get some quick wins, prove ROI, basically market that with your teams and say, you know, congratulate people if they’re being creative in how they’re using AI in their day to day.
And just build on, on the early wins. And I think, you know, as, as this comes along, AI is so new, you know, it’s a, it’s a space where we’re all gonna be in this together, kind of learning as we go. But it should be a really exciting time. It should be a transformative time for our, our industry and, you know, society in general.
Lori Boyer 30:18
Fantastic. Well said. AI’s awesome. Give it a try. Just do something. Don’t, don’t let, you know, as, as Tom said, don’t let perfect get in the way of better. So love that. If you are using some great AI tools, I would love for you to share. You can throw ’em in the comments if you’re watching on YouTube or if you’re on social media.
Feel free to email if you have questions around AI tools. As Tom said, we’re all in this together, so let’s work together to figure out what you know is working for you. Tom, thanks again. Thanks for speaking with me at Parcel Forum. Thanks for being here and just being such a great supporter of the podcast, and it’s been wonderful having you.
Tom Butt 31:03
Yeah, always a pleasure, Lori. And excited to see what, what people come back to us with. So appreciate it.
Lori Boyer 31:10
Awesome. We’ll see you all next time.
