.png)
The Doing Business in Bentonville Podcast
To create an ecosystem that connects leaders of all kinds – industry, community, student, educational, civic, investment and entrepreneurial – to help overcome Omnichannel Retail barriers through exclusive, insight-rich content.
The Doing Business in Bentonville Podcast
Ep. 118 - AI That Solves Supply Chain Chaos
What happens when warehouse chaos meets artificial intelligence? Sean McCarthy, co-founder and CEO of BackOps.ai, joins the Warehouse Whisker Warriors to reveal how agentic AI is transforming supply chain communication and problem resolution.
Drawing from his experience at Amazon Shipping, McCarthy identified a critical gap in warehouse operations: while systems existed to capture issues, humans still had to manually gather information across multiple platforms to solve problems. BackOps' solution utilizes an agentic AI framework that operates concurrently across systems, rather than following the linear paths of traditional Robotic Process Automation (RPA).
The magic happens when a warehouse issue arises, missing pallets, damaged goods, shipment discrepancies, and the AI simultaneously checks carrier systems, vendor portals, warehouse management systems, and even contacts stakeholders. This parallel processing dramatically accelerates resolution times while maintaining accuracy and working through existing communication channels like phone, email, or messaging apps.
Most impressively, the AI sounds remarkably human. During demonstrations, even experienced professionals have mistaken the BackOps AI for a human agent. This natural communication style helps overcome the hesitation many feel toward AI systems, especially in relationship-driven industries like supply chain management.
For companies concerned about removing humans entirely, BackOps offers flexible human-in-the-loop options while providing measurable ROI from day one. The system also builds predictive analytics capabilities, helping identify recurring issues with specific carriers or vendors before they become problems.
Ready to see how AI can transform your warehouse operations? Contact Sean directly at sean@backops.ai or visit www.backops.ai to schedule a demo and experience firsthand how their system is creating the future of intelligent supply chain resolution.
Welcome to Doing Business in Bentonville, the Warehouse Whiskerid Warriors. I am your host, josh Safran, from the studio in Bentonville, and my co-host is the well-whiskerid warrior himself, mr Williams, with another W. How are you, harvey?
Speaker 2:Hello again, hello, hello. Over the 4th of July holiday weekend, I was told that my beard was getting too long by my mother-in-law, which Shout out to your mother-in-law, Harvey.
Speaker 1:We don't normally include her in the podcast, but that's exciting times.
Speaker 2:No, but it was a low point for the weekend for sure.
Speaker 1:Well, if that was the lowest point of the weekend, then you had a good holiday. That's for certain. Well, if that was the lowest point of the weekend, then you had a good holiday.
Speaker 2:That's for certain. Yeah, but we're here for backups, yes.
Speaker 3:Sean Guys, thanks for having me. Yeah, I feel left out. I didn't know that the prerequisite was that you had to have a great beard to come on, you guys, you forgot to mention that.
Speaker 1:Well, Sean Elmrob, who's very familiar with the podcast, former plug and play, current executive over with Back Ops probably should have given you a heads up. We will probably take many jabs at him today. This is the first one that he did not do a good job of prepping you for beard looks. Maybe you should have taped something on or bought a mask or something if you couldn't grow it. But Sean did not do a good job of getting you set up for today, so we'll have to give him that feedback.
Speaker 3:Yeah, do a good job of getting you set up for today, so we'll have to give him that feedback. Yeah, yeah, I'll jot it down. I'll get a sticky note here, and maybe it's one of many.
Speaker 2:You know, we do really well during playoffs, even because everyone's naturally sporting a playoff beard. But because we're in the summer dead zone, I'm listening to like fantasy football podcasts Like it's dark, it's dark. We got no sports happening right now.
Speaker 1:When we have the NFL kicking off soon. Right, I mean we have baseball. It's hard to watch baseball in 110 degree heat, so it's not a great season for sports. Yeah, and more importantly, we are back to Sean McCarthy. Can I call you the CEO and founder? I get in trouble if I miss people's title up.
Speaker 3:No, yeah, that's our co-founder and CEO. It's just fun. No, yeah, that's our co-founder and CEO. It's just fun.
Speaker 1:Perfect, and can you give us the elevator pitch on backups? We've had some really cool, interesting tech on here and I've had a chance to spend some time with you the last month or so out in Silicon Valley and what you guys are doing is super cool, very interesting, very CPG supplier retailer relevant. So we'd love to get some time with you today on the Bentonville podcast.
Speaker 3:Yeah, yeah, no, I'm happy to. The quick high level is that we noticed there were a lot of systems out there that could catch all of the inbound inquiries or issues that come into these warehouses and open up a ticket, but a human is still having to go get the information or fix the ticket or chase down the problem, and with AI they don't necessarily have to jump through those hoops. So we built a system that can intake those problems and help a human work a little bit more efficiently and obviously resolve the issues in a faster manner and give the end customer the best experience.
Speaker 1:So it sounds like another AI company, harvey, right? I mean, it's just super on trend with everything going on, and there are companies that say they do AI and then there's companies that actually do AI, and Sean and BackOps are very, very heavy AI and do a fantastic job in this space.
Speaker 2:Yeah, and actually maybe Sean too, this would be helpful. Tell the listeners a little bit about your background coming from Amazon, right, what you're building, what you learned at Amazon and kind of what and how that translates to today. And then, yeah, as Josh mentioned, I mean I think it'd be really helpful to get a little bit of your perspective on the lay of the land when it comes to AI, when it comes to freight tech and kind of the world that we're living in today.
Speaker 3:Yeah, I'd be happy to freight tech and kind of the world that we're living in today? Yeah, I'd be happy to. So the last few years before this endeavor, I was at Amazon, initially hired there as a global sales leader in the marketplace and had an opportunity to go to Amazon Shipping. If you're not familiar, amazon Shipping was created to tackle a lot of that ground parcel that one to 50 pounds and initially via the shippers and folks that were already using Amazon. But now it's expanded past that. So I spent a lot of time with our customers in the warehouse just really understanding their pain points, and that was where I saw a lot of the inefficiencies and essentially what's coming in on the day-to-day is these different issues or inquiries that are a different flavor of the same thing. So there's high repetition in those workflows and so not to be another AI company, but this was purpose built, with the knowledge that I got at Amazon.
Speaker 3:In working with these warehouse owners and operation specialists and the people that are on the ground doing this, we focus on one thing and this is it, and I think that that's helped to where, when we are working with our customers, we know already a lot of the issues that they're experiencing and we're able to showcase how our solution already helps with those things.
Speaker 3:It's not start from scratch and give me all your issues. It's hey, we know you're experiencing a lot of this and this is how we are helping, and so I think that helps for customers to conceptualize the AI out the gate and really understand like, what is it going to look like from an impact perspective? Because it's already kind of one, this like black box of AI, like it can kind of do anything, and so we try to make sure that it's something that is pretty focused and purpose built, you know, at the beginning, so that they can measure ROI. Like, hey, is this something? And, unlike a lot of legacy SaaS, the day you turn this on, you see ROI, it's super tangible. So if you are looking to measure from a vendor perspective, is this good or bad? You're going to get it right away, you don't need to have it for six months, and that's something that we pride ourselves on.
Speaker 1:Sean, let me ask you a question, because we're here in Bentonville and I worked in CPG space for a while and when I talked to you or I met with you and I started understanding the technology, I go oh, this is a huge miss or huge opportunity for companies. So, in layman's terms, for me, if I am CPGA salesperson, replenishment person that is on with Walmart and Walmart is expecting a truck of my product, and the replenishment person at Walmart then calls our team back and goes hey, the truck's late, the truck arrived. We're expecting a full truck, half truck, 15 pallets, five pallets a thousand cases and some of it's damaged, some of it's missing, some of it's not there. The panic that would ensue on the supplier side was crazy and it would take multiple phone calls, all kinds of human interaction, days before we can get a response back and figure out what the heck actually happened. You've solved for all that, correct.
Speaker 3:Yeah, yeah, we have, and we see that's obviously a very common scenario where you have something and what we focus on there are what are all the things that a human would normally have to do to go resolve that. And so our system can be triggered with a number of different ways. So it can be done with a phone call or a WhatsApp message or a Slack message or an email. Essentially, any medium from a communications perspective can trigger that. And so where that becomes really powerful is in the scenario you just gave. Where there's supposed to be a certain number of pallets, I can be an operations manager in the warehouse. I can be at the bay and make the first phone call in and say, hey, I just received this truck, there's supposed to be X number of pallets, or there's supposed to be X number of items and there's something's damaged or there's something missing or whatever the problem is. And so what it will do on the backside of that is we've predefined what the standard process is for that business, what a human normally does, and so the AI will just start to work down that same list. So it can go into carrier systems, it can go into vendor systems, go to warehouse management or order management. The difference is that it's an agentic framework. So if you're familiar with RPA or robotics process automation, you know it has to check the first box and then go check the second one and then go check the third, like a human would do. With an agentic approach, we're obviously deploying agents, and so those agents are going to check all of those things at the same time and bringing it back right. So it could be checking a carrier or a vendor portal at the same time as it's calling Josh and slacking Harvey to aggregate that information to get the fastest resolution. And so where it pulls ahead is that the AI is pulling kind of that orchestration string to be that central system and times that now by many issues let's say that these are concurrent and it's happening at multiple warehouses at the same time.
Speaker 3:You're starting to get issue resolution, but what comes after that is actually the most important thing, which is the predictive analytics as well. You're getting all this data. Now what do you do with that data? How do you make business-driven decisions? Are you having a carrier problem that's constantly missing something or damaging something, or is it a vendor problem?
Speaker 3:And where it pulls ahead, especially on CBG, I know a lot of times where you might have outsourced 3PLs as an example, with really no visibility into those 3PLs because there's multiple WMSs or different order management systems. And this is the first time in history that you can kind of leapfrog all the integration problems with just simple communication tools that we already use. And so that's really the vision for this, the Relay product that we've built. It becomes something much greater. But when we go back to just tangible results, it's great the first time that that shows up. You have a palette mismatch or a BOL mismatch or a shortage and you just have the AI take. It Is human, still in the loop, in the sense that a lot of those communications back and forth to the customer can still be within your realm if you want to own that, but all the legwork is done by the AI.
Speaker 2:I'm glad you mentioned RPA. We were hosting a CIO dinner recently and we asked the question what technology investment has produced the highest ROI for you over the last two to three years? Vanilla question not quite sure what you're going to expect because we had a lot of brands there. We had uh, logistics providers there, transportation carriers, etc. Um and uh. I think it was like nine out of ten respondents came back and said rpa like for them, rpa created roi.
Speaker 2:Now, which is kind of um, it was eye-opening to us because rpa was really hot in in the early stage in like 2017, 2018, 2019. Right, that took almost six or seven years for it to finally get through to a deployment, rollout and roi producing numbers to be able to validate like this investment made sense. When you're going and having conversations around backups, are you, are you going to be working with an RPA, removing an RPA? You know like, how are those conversations today when you're saying, hey, this is, this is the next best solution version of RPA. What you guys have is outdated and legacy. I'm curious. I'd love to get your thoughts there, because I know a lot of people are happy with that ROI and I think what you built at Backoffice is amazing and that's going to hopefully create even more ROI for people that want to adopt it now.
Speaker 3:Yeah, I think it's a good question, and obviously it's business dependent based on what they've got already in place. If they have RPA type tools or they've outsourced, that it actually makes it a little bit easier because they've already conceptualized the basis of the foundation of what we're trying to accomplish. But we can just do it in a more cost-efficient and faster manner with an agentic approach. I think where RPA lacks, though, it's a lot easier for us to come in and showcase some of the ability that, as an example, when we're collecting all of those phone calls and emails, slack messages, and you want to start to query against that information, it's much more complicated to do some of those steps with RPA. So I think that it depends the short, but what we do find is it's definitely like a I don't want to call it a rip and replace, but it's.
Speaker 3:It's a replace solution for sure, because obviously it'd be repetitive, and with that we are able to articulate like current state solution versus future state solution. So we'll map out okay, this is a specific process so they can conceptualize it so like, let's talk about, you know, claim and reshipments or something like very specific in how they have the RPA solution set up, and then we can measure time to resolution. On that we can show time to resolution, accuracy, visibility of data and like different key metrics of what the new solution is. And the nice thing about it is, if you look at a cost perspective, like, even though it's a vastly better solution, um many times that I see it it could even be more cost effective to switch to this. You know, with the way that that costs are driving down, so it's not like this huge burden that we have to say, okay, when I spend double it's and it hit that it's either at that same cost or about that same cost or even even less for a solution. So to that extent it also helps a little bit.
Speaker 1:Sean, two-part question from my end. So are you able to quantify man hours saved and or time spared in doing a project like this compared to a legacy way somebody's doing it and utilizing your tool? And then, second part of the question would be if I'm not quite ready to make the jump yet to full agentic AI and remove all the humans here, this still works with human in the loop. There is still a need or a want to have humans around. Just kind of answer both of those questions if you wouldn't mind.
Speaker 3:Yeah, so I'll start with the back half first, so you absolutely can keep a human in the loop. We actually recommend, when you begin to keep a human in the loop and the system's dynamic enough to where you can say, hey, this customer is really important to us. If somebody from this organization reaches out, no AI touching it, this will be straight to a human in the loop. Or, even more specifically, if this contact the president of this company, who's our main stakeholder reaches out straight to a human in the loop and so you are able to, I would say, kind of make sure that that trust is built and it's pretty dynamic. You'll have to tell me the first part of your question. I apologize.
Speaker 1:The first question was are you able to quantify, compared to a legacy system versus today, how many man hours, how much time you're saving, how fast you're finding solutions, as compared to any numbers or anything that you could share that would be able to quantify some of the work you're doing?
Speaker 3:Yeah, absolutely, so, absolutely. Everything that we do is measurable, which is the other key thing, and so when we build these dashboards, these dashboards are custom built for what's important to the business, right? So is it speed or is it the accuracy? We have a customer that is using this to generate quotes, but the quotes are usually 10 to 12 steps of process where they're logging into multiple vendor portals. If the pricing is more than two years old and the ERP has to go fetch new pricing, there's a number of things that a human would normally have to do. So for them, it's accuracy making sure that it's at least on par with what the humans are doing, or if not better, and then time to resolution.
Speaker 3:When you have an actual inbound come in with a problem, what that looks like, and so those are usually like speed, accuracy and then kind of that quality are like the top three buckets that we usually see measured, but you have full visibility to that right, everything from that initial call, whether that's an internal operations person making that call with, like the pallet problem we talked about, or a customer that's making a call saying like the pallet problem we talked about, or a customer that's making a call saying, hey, I have a problem. All of that is recorded, logged, and so you can start to track from that point all the way to resolution, so you have full visibility across the spectrum. And where that also becomes important is like we're ingesting that SOP, right, or that general process. There might be an inefficiency in your process as well, and now you start to get visibility on that as well, where maybe it's taking longer and you can start to look into these things and identify like, oh, we have a step in there that is unneeded as an example, and so it's highly measurable.
Speaker 3:And that's how we usually start with customers too is like we'll start with one or two process. That's usually pretty human, intensive. Just start there. We'll ingest those SOPs, we'll show you how it works. You're going to tell us what's most important to be measured for success, like how do you define success within each one of those realms? And then from there we start to expand to other items and everything's in a dash. And you can even say, hey, have this report to me every Monday at 9am as an example. You know, with this criteria. So I think that's also part of the reason why the solution works, is just that like there's nowhere to hide, right, it's like very, very blocking way.
Speaker 2:Now as the founder, how do you stay focused, which is maybe an internal question, but how do you keep your customer focused right, Because they're going to look at this as a magic wand, and they have been ever, since. This was more like a machine learning problem a few years ago. How do you keep your customer focused and not allow for mission creep Because they're going to say this was awesome, Can we go do X, Y, Z, you know ABC. You're going to say slow down without the brakes. How do you handle that as the leader in the organization?
Speaker 3:That's a great question, I think, for us we want to be working on problems that are painful to the organization. Right, and I always like use a scenario with even our folks internally, where you know you talk to a customer and they give you a scenario and it's like great, you know, from a one to 10, one being like doesn't affect you at all to 10 being like. When this happens, like it's an outage, the whole thing goes down and it's critical to the organization. Where is this process or you process, or where does this sit? And if they say, oh, it's a two or a three, it happens once or twice a month, those are not the right problems that we want to be working on. We want something that's an eight. It happens all the time, it causes so-and-so. To go through this, it's super painful for us and it's high impact. Obviously, it's something that they hold near and dear from an organizational perspective, but that means we're working on the right, the right problems and so, to answer your question, that happens a lot where they're like oh well, we also have this, this finance report we have to run Great Well, tell me about it. Oh well, we have to run a finance report once a month and it takes an hour. I hate doing it Like, no, that's not something, not something. Yes, the ai for sure can solve that, but that's not a good use of it.
Speaker 3:You know, and I think what we're working on is the ability and very soon, uh, you'll have the ability to use natural language to uh describe something or just record. You know, one thing that we found is nobody wants to make an sop right, like, uh, typing up an sop of anything just sounds terrible, and so we did build something to where it pops up video, records the screen and you can talk through the process. It extracts that out and automatically creates an SOP. So you can use that, or you can just upload a standard SOP or use natural language to start kicking these things off and deploying them, and so we will put that control into the customer's hands very soon. Where they want to do those things, great.
Speaker 3:But for right now, you know, we try to keep it focused to things that are going to be like, really important to the business mission critical, um, cause that's the best way. And then highly repetitive tasks, right, where it's happening daily, uh, or every couple of days at least, so they have enough data to measure. If it happens once a month, then we can measure it in like six months, because then you've got, you know it happens six times, right. But if it happens multiple times a day, by the end of the, if we deploy on Monday, by Friday, you're like hey, this is awesome or I'd like it to do this. That's different. We can move at that same velocity.
Speaker 1:Sean, if I'm sitting here as a local supply chain lead dealing with Walmart, Sam's Club et cetera, and I'm thinking this is pretty cool, but I'm not the decision maker, I need to funnel it into my corporate organization. Who's the right point of contact? Is it a transportation lead? Is it a warehousing lead? Is it a supply chain innovation lead? Who should I be getting on the phone saying?
Speaker 3:hey, you got to meet Sean and the backups team. I would say some of the folks that you named we're usually integrated with, but it's, you know, vp of transportation, director of operations, director of innovation or VP of innovation. Those are probably like our usual folks that we talk to. I think one thing that's very interesting and that we try to incorporate off the bat is, like, when you talk about who is actually being, who are the first people that are impacted and are going to benefit from this? It's usually like those operations level folks right that are having to go jump through.
Speaker 3:The vp of transportation or you know, is not necessarily getting on the phone chasing these things down, and so we do try to include those folks, to just levy up like, hey, how big of a pain point is this really? Like, like, how much of your day, of your, of your eight hours of the day is is spent chasing these things down, and what we find is like that's very, very impactful, because there there isn't always direct visibility to that same extent, though. Uh, as our product suite migrates and as we build into it, this roadmap, uh, we actually are building it for more of the executive suite as well. That increases their visibility and it actually serves them in a different manner, and so there's kind of buying across the stack. But those are usually the folks that we work with.
Speaker 2:What's coming down the pipeline for you guys on the roadmap side of things? What are you cooking up in the lab right now for the customer?
Speaker 3:There's a lot, so what I'll share is you know, I think there's a few different things to highlight. So, first and foremost, we have a new product that's launching here in a few days called Doc. So we're really excited about that, and from a high level. Doc lives in the communication tools that you already use. So you know email and messaging systems like Slack, where, as inquiries come in, it will automatically read what that is. It'll auto go, fetch it out of your order management or warehouse management system, return the answer and you can modify it and just click send or you can say, hey, I want more information. It also has read-write ability, but it lives in your communication, which is just a huge unlock for speed, speed execution there. So that'll be launching here in the next few days.
Speaker 3:Further than that, really, what we believe here is, you know, the relay system, and relays is the system that we've been talking about, that we built. It serves a lot of practice and what we notice is there's usually kind of the way I think about it is like you have like kind of the relay system at the top of the funnel and all the way down at the bottom you have like the core systems of record, like the warehouse management system, the order management system, tms, etc. And in between you've got all these SaaS stacks right, ticketing systems and project management systems. You can even maybe throw like a CRM in there that's housing customer data and a lot of that's already built into our system to where you know you don't have to use an external ticketing system, you can use ours. We have a project management type system already included. You know you can obviously house all of your customer data.
Speaker 3:Any SOPs or any documentation can be housed, and so we start to look at more of a unified stack of like the central nervous system that you can use, because, again, we're aggregating everything across the stack and so we're building the components into that when kind of launching those strategically. So that's probably what I can share on the roadmap, but we're really excited about it and everything is purpose built again for our industry, with the voice of the customer in mind, just with our problems that we face, and I think that's something that's important and I think that, with all the AI madness and the tools like you guys kind of mentioned, you do need something that's deeply vertical in terms of you know it is built for the problems that we face because everything's different and I think you can get to a certain extent with some of like the mainframe tools out there. But at some point, you know, when you're logging into vendor systems and carrier portals and getting information out of multiple OMS, you kind of have to have some domain expertise and we hope to be that go-to.
Speaker 1:So, sean, I've been on multiple podcasts with Harvey. He was hoping for a drumroll and you were going to break some news right here on the podcast, but obviously you didn't fall for it and by knowing going to be viewed by millions, uh, you didn't want to want to come out to your own.
Speaker 2:We'll take shout out for a couple of beers later. I'll get it, I'll get. I'll get the secrets out of them at some other way, yeah, yeah. Yeah, it's okay. Look, I'm not deterred by any stretch.
Speaker 1:I was going to make a comment that you can let me know if you want to layer onto it. I've had the pleasure of sitting in a couple of meetings with you with some leads where you actually demo their product, and it's not often the guy that's the lead looks over me like shaking his head. Often the guy that's the lead looks over me like shaking his head because when you get on the phone on the demo and you're talking to an AI agent, when the call was done, he leaned over in the room and said that was a human right and I was like I looked at you because I was like I don't know, I think it was an AI agent, but you're like no, no, that wasn't Get out of here. There's no way that wasn't a human. It was that impressive.
Speaker 1:I know you don't want to do a demo here and I know obviously we would give your email address at the end and get people to reach out because the demo was super impressive. But anything you could share or comment, the demo was just. I was blown away and how articulate the AI agent was was insane.
Speaker 3:Yeah, no, thank you. I appreciate that. I think for us, in the way that we're trying to build the product, there's still an inherent hesitation to utilize AI and to, more importantly, like, roll it to like. For our customers to roll it to their customers I mean, that's the lifeline of their business and I think in supply chain, as like a broader ecosystem, it's relationship based heavily and if you're going to roll something out, it needs to be. It is okay to say, hey, it's AI, but it needs to be something that is seamless. That isn't this like massive disruption to how things are done, where, like you're used to talking to a human and all of a sudden you talk to like this robotic, you're probably going to go a little nuts.
Speaker 3:And so, to the extent that we formatted the voice to do the uhs and the ums and really sound like a human, and I think that the voice, not just from us but from the broader ecosystem, it's getting good enough to where, eventually, people will have a tough time disseminating if they're talking to a human or an AI. But I actually think that's not the important part For me. If it's 8.30 PM on Thursday and I don't know, I'm trying to get a flight change or I need something done. I don't mind talking to the AI, I just care about what's going to get me the resolution fastest, like change of flight or book the. Whatever I'm trying to do, just do it quickly.
Speaker 3:And so there's two ways in the way that we've established the product and kind of what you saw in the demo is number one, from the front, facing like let's make it sound like somebody that you'd want to talk to and that's friendly and kind of sounds at least like a human and not you know, you don't have to deal with language barrier or you don't have to deal with their kind of robotic voice. And then number two, like what's the fastest way to get you that resolution? And some of those things can't be done right away, but you're going to leave with some sort of closure. Whether that's a new ticket, id or resolution, you know things are going to get solved in a faster manner for you, and so that whole kind of streamlined process is something that we've included in the product and that's what you heard in the demo, but I would love to give anybody a demo and walk through it live.
Speaker 1:And is it fair to say, though, you may be thinking something and I may be wrong, but, like, if you're working the traditional, old legacy manner, you may not get somebody to answer your call at three in the morning, at 10 o'clock at night. You're going to go through the AI system, right? You're 24-7, no.
Speaker 2:Yeah, absolutely so. Ai doesn't need breaks, you know, and it's 24 think that's not yet.
Speaker 3:You know, maybe maybe some regulation comes up. Yeah, yeah, they're all working for the regulation where, uh, that's my given, given our ai breaks. But yeah, I mean, you're spot on, I think. And that's the thing, like in our industry, like issues happen all the time and so if it's, if it is, you know, uh, a driver has a problem in the middle of the morning or late at night or whatever it is, there's fastest resolution where some of that can be solved completely autonomously. Maybe it's like a rebooking or it's a reordering or change order or something that the system can just do on its own. But even if a human has to get involved, at 8am when that human comes into the office, they've got all this information.
Speaker 3:Hey, sean, here's what happened. Josh called in. It was two o'clock in the morning. You can listen to the call. You can look at the call transcript. You can look at everything that the AI did. She said Josh called in. Here's this problem. Here's everything that we found about the issue. How do you want to handle this? And you can start to train. Say okay, next time it does this, I want you to do X.
Speaker 3:So it helps to train the LLM or just simply you've got all this ammo now at 8am in background to make a quick decision, and so that time to resolution again just continues to shrink.
Speaker 2:We are in a wild time because, like freight tech is maybe 10 years old, 15 years old, think about like Project 44 and the big guys right, like they got started right Maybe 15 years ago.
Speaker 2:We've gone through a few waves since then, and the most recent wave is the one that you guys are currently, you know, one of the leaders in which is you have an AI tool that started to take over or eat or replace business logic, like in a lot of ways, like building vertical status applications anymore might not be able to compete with what backups is building today, because of anymore might not be able to compete with what BackOps is building today, because of your ability to be lean and flexible and to like move into different work streams in a really unique way, which I think is fascinating.
Speaker 2:I also think it can probably prompt a little bit of hesitation from the buyer side, as we kind of talked about a little bit earlier, right, which is like, wow, you know you're making people kind of second guess themselves. I do think it is a benefit, though, that people have had a few years now of working with ChatGPT and they can kind of understand it and they can visualize it and see it when you have these conversations, sean, like what's been the unlock for you in trying to get people to see the world the way that you see the world.
Speaker 3:I think there's a few different components. First, there's not one singular system as it relates to, like this information aggregate that has everything that a human's going to want to need, especially in our world, right, like you are going to have to contact carriers and vendors and go through a WMS or TMS or LMS in multiple systems to either find the answers or get a resolution. And so I think for us again, even if they have all of these systems day one, it's not a rip and replace, right? So let's say there's eight systems that accesses for an issue that comes in. We sit on top of that at the application layer, so we just tap into everything and you can access all of that data via a normal communication tool like an email or Slack or you know, using our system. And so that's usually something that they're at least like willing to look at and see like, okay, well, if that's possible, it's great to have that information.
Speaker 3:But then what's the next step? And for us there's like kind of two channels, like you have like the read how do I read all that information? Out of that? I also want to write. I need to be able to update the system because I don't want to have to still go back and do the updates myself.
Speaker 3:It should be able to do both on a lot of these, and so with that, usually we're taking like, okay, let's just start with one process. We'll show you end to end and we prototype it, start to finish this exactly how it would look in your system, and you get exactly what we prototype for you, and so that's something that I think is super helpful, because you start to understand the end-to-end process and then you also start to understand like what it would look like in like the real world, and so that's usually our approach. Sometimes people have seen like other systems where they have something in their mind or maybe it's like an idea of like it would be really cool if it could do this. But to your earlier point, where we'll be very like upfront, whether you have to, can, maybe you shouldn't do that, or know that the technology is not quite there yet.
Speaker 2:As a founder in this space backups maybe not the center of this question. What are some AI traps that you can kind of share with the listeners that people may know? Maybe they don't know, but being you know in this, this space, you can kind of see how the competitors are doing it. Stuff that has not worked for backups in the past has not worked for your customers in the past.
Speaker 3:Is there anything that you can share that that would be kind of a bit of a landmine of like hey, yeah, if you hear this, ask questions yeah, I think that the big thing there is the AI only knows what you give it right, and so if you go to ChatGPT and you start asking history questions, it's been trained on that context so it can give you the answer.
Speaker 3:In the business world a lot of times people are very quick to say like, oh, you're going to turn it on, it's just going to know everything and do these things, and that's actually not the case, and the reason why is that a lot of times humans have the knowledge in their heads right Of like specific business process or the way that we handle this one vendor, or the way that you know this certain account is set up or the contact, and so that a lot of times is not written anywhere, it's not in a database, and so the AI wouldn't know that, and so it is something that when you get into this you're helping.
Speaker 3:You can dump documents in there, but AI will start to ask questions to you as well to learn, and you have to give it that context.
Speaker 3:So that part does take a little bit to continue, to build it to where it's like something that you know is know your business and can actually make really tailored recommendations, because it's been given a lot of context. So the things to like be wary of is if they say like at day one, you're going to turn on, it's going to know everything, it's going to do these process and to end like it's actually not necessarily the case. It depends, right. If we've got all that context up front where you know you use our SOP builder verbally, you said okay, well, if this happens and the vendor's not there, then you're going to do this and this. We have all of that information, then of course, it'll have a more success rate, but it only knows what you give it basically. And so that's something that our new product that we're launching later this year will help with, because they'll have more context, but for right now you still have to give it the background so that it can finish the jobs.
Speaker 1:Sean. I want to go backwards to Harvey's previous question about some of the barriers and again I'm just thinking about this logically the existing. I don't want to work with Sean because I don't know AI. It's so new, I don't trust it. I got a team of people in place and they've been doing it for forever, so that could be a barrier. But I'm looking at it very logically from the other end.
Speaker 1:There's a problem that's happening somewhere along the way and they make a phone call in and they call Harvey. Harvey doesn't answer the phone. Harvey's out on a bathroom break, harvey's out at lunch. Harvey answers the phone and Harvey says I got 15 other things more pressing and I'll have to get back to you later. They email Harvey and it sits to the bottom of Harvey's inbox All problems that are real in this world. Then they happen to get Harvey With me in particular. Well, I'm going to get to me in a second. They get to Harvey and Harvey's immediately able to help and says I have to pick up the phone now and call Josh. Josh is on vacation. Josh was terminated last week. Josh is also too busy.
Speaker 2:There's like a million Explain the Hawaiian shirt that you're rocking right now.
Speaker 1:Thank you. It's my Nike shirt, thank you, but the point is there's so many places where I just caused problems by interacting directly with a human that may not think it's as important as it needs to be, versus I'm going to try AI and none of those things are going to happen, because it's not going to take a bathroom break and it's going to move quickly and move through the system. So it almost makes sense to go to a pilot with you just to test this thing out. I can't understand why there would be more of a reason not to want to try to have a conversation with you Just based on thinking logically around. Harvey and Josh are going to struggle. Potentially, the AI will have maybe different struggles, but it's going to move a lot faster than these two yahoos.
Speaker 3:Yeah, yeah, and the pilots are great. It's a really light touch way and, again, we keep them hyper-focused. Maybe just one or two processes during the pilot, just to make sure that everybody's on the same page. You can conceptualize it, and then obviously it starts to get the wheels turning about the other possibilities. The nice thing is I talked about the human knowledge is in the head, and sometimes AI will need to call a human, and it obviously can do that without an issue. What we do is we set up so you don't have single points of failure, though as well, so it can email you first, josh, and you don't answer. So it tries to call you an hour later and you don't answer. So then it goes to Harvey and he doesn't answer, and so it can start to move up the stack and do its own reasoning as well.
Speaker 3:The problem with the humans, when you're tracing down, is like, oh, I'll do it tomorrow, or I just give up, or whatever it might be, and so you do need somebody that stays on top of it. We see that a lot with claims Claims is huge in our industry, right, where it's like you have to wait two weeks to file and then the human just forgets, right, or they do. They file it and then it gets denied and they're like well, I'm not doing the appeals process, it's so much work, I'm not doing that Same thing. Ai will stay on top of the pilot. We will measure, like what's the criteria for success?
Speaker 3:Again, like for the, for this process, what's the current state of how it's being done now? What would you like it to be? And a lot of times, too, we see that the humans are very much still in the loop. But it's just all of the coordination and efforts and manual research and grabbing these informations no longer has to be done. They're equipped with that right out the gate and obviously the end customer is getting that resolution. So, yeah, I totally agree.
Speaker 1:Harvey, one final question from you, because I always want to leave it with the best and the brightest to tackle, as opposed to myself. And then Sean will ask for LinkedIn email address website best way to connect you for a demo. But Harvey's got something brilliant. Maybe he'll come back to see if we can get some more insider information from you on some new product.
Speaker 2:But Harvey always ends with something robust. I'll take a slightly different approach. It's a funny job market today. Right, we're trying to hire AI engineers. It's quite difficult to do. Do you have open roles right now? Are you looking for key hires right now? Are you looking for people either on the sales side as well as the engineering side? How are you attacking growing this company from the team perspective, because I'm sure there are a lot of people that are listening that might have a network of folks that are ready to go to help back ops continue to grow.
Speaker 1:And you're not looking for a job here, Harvey, just to make sure we're all clear You're not throwing your name in that hat because that was-.
Speaker 2:I saw the Hawaiian shit and I thought maybe I could throw Josh a bone.
Speaker 3:Yeah, yeah, no. I mean yeah, we are hiring for engineers, we're actually hiring in person and we are hiring some sales folks as well. That's starting to come, so you know we can talk about that and the folks obviously that we're hiring from a sales perspective, looking for people that have sold into this industry before and understand it, because it's not just the typical SaaS sale that it was back, but we're looking for engineers. Funny story for you guys, really quickly, is there are a lot of cheating tools out there now with AI and these engineers are-.
Speaker 2:For the interviews.
Speaker 3:They're cheating, they're using the tools, and so we actually have to have them come in now and do their coding interviews in person so that we can monitor them. So we are looking for in-person engineers. You know, to be sales folks not as much, but yes, it's hilarious.
Speaker 2:What was your wake up call of like oh, this was too quick.
Speaker 3:You know. So we have like a pretty defined process of like how we hire these, these engineers, and, uh, multiple people were just acing it like acing every single coding question. I mean like genius level is is what it was told to me. And, um, you know, we try to stay up on what's happening, uh, and then we had one person you have to come in still for one of the onsites and it didn't quite match up. I talked to another peer of mine and they're experiencing the same problem. They're doing the same thing having people come in. It's the new world we live in. That's the one good thing I guess about sales live in. But that's the one good thing I guess about sales. You can't really use an AI tool on that Very relationship-based and you're either going to sell it or not.
Speaker 1:So yeah, sean, in closing, first of all, again, fantastic company. When we spend time with you out in Silicon Valley sitting in the room with companies looking for solutions and then head nodding and blown away with you speaks volumes. So anybody that has not had a chance to work with backups meet you and team, meet Sean 2.0. Let's give them email address, linkedin, ways to connect with you. Website, because you should get a lot of new business and a lot of new leads Because, again, what you guys are working on is fascinating.
Speaker 3:Thanks, josh. I appreciate that. My email and anybody can email me at any time Fastest way to get a demo as well is Sean S-E-A-N at backopsai Website is backopsai and I'm on LinkedIn. I think my username is Sean B McCarthy. I would love to connect and chat and obviously we can showcase some of the unique things that we've built, kind of pervertical or more unique to a business. So we welcome all of the opportunity there.
Speaker 1:Yeah, the demo is outstanding. So hearing the podcast today is great, but getting some time with you and actually seeing the tool is even more impressive in person or virtual, so I would highly recommend folks get access to you to do that.
Speaker 3:Awesome Thanks, josh, I appreciate that.
Speaker 1:Harvey, any closing thoughts about Backup Sean the world that we're living in? I always like to throw it.
Speaker 2:Shout out to the other Sean. You know Sean Elmerab. You know if anyone's for demo you might get Sean Tudato. Or you know people call him little sean sean jr.
Speaker 3:I've heard a lot of different ways to refer to mirab mini me exactly it's mini me, but he's, you know, I don't know sean's, he's very, very tall, he's not, he's not many, but I'll tell you, hiring, hiring, somebody with the same name, same spelling, you know, has has presented its own. Uh, but we're, we're thrilled to have within, you know, it's been nothing short of an adventure.
Speaker 1:Thanks for making the time for us today, you.