我创立了RoboBuy,一个全球性的供应链分析平台。我们专注于帮助大型企业优化其供应链管理。我们的工作流程是从客户的各种ERP系统(有些企业甚至拥有40多个不同的系统!)中提取数据,然后利用自主研发的AI算法对这些通常“脏乱差”的数据进行清洗和分类,最终将其转化为清晰易懂的洞察和可视化仪表板。这些信息能够帮助首席采购官、财务主管以及风险管理人员做出更明智的决策,例如识别潜在的成本节约机会、优化支付流程,以及识别和规避供应链中的风险(例如制裁、现代奴役和网络安全风险)。我们甚至可以帮助客户分析其在少数族裔拥有的企业上的支出情况。
生成式AI的出现,彻底改变了我们的行业。起初,客户对像ChatGPT这样的工具能否替代我们专有的AI算法感到疑惑。事实证明,这些工具并不能直接替代我们现有的功能。然而,生成式AI的快速发展速度令人叹为观止,它正在以前所未有的速度重塑着各个组织和行业。未来几年,我们将见证自动化程度的显著提升。
生成式AI为供应链带来了前所未有的机遇。例如,它能够帮助企业更深入地了解其多层级的供应链,甚至可以追溯到供应商的供应商,从而识别潜在的风险。此外,生成式AI还能显著增强预测分析和情景规划能力,帮助企业更好地应对各种不确定性。大型语言模型(LLM)能够处理非结构化数据,这对于全面了解供应链至关重要。
Meta和苹果等科技巨头纷纷进军AI人形机器人领域,这将进一步加速物流和供应链的自动化进程。然而,现有的ERP系统,例如SAP和Oracle,其代码库往往非常陈旧,难以快速适应AI技术的变革。我认为,未来将涌现更多基于生成式AI的纯AI玩家,它们可能会对这些传统巨头的地位构成挑战。
供应链管理中一个长期存在的问题是数据质量差。许多企业仍然依赖人工数据录入,这不可避免地会导致错误和低效率。生成式AI能够自动化这些流程,显著提高数据质量。此外,它还能增强预测分析和情景规划能力,帮助企业更好地管理风险。
生成式AI技术正变得越来越普及和易用,即使是非技术人员也能轻松上手。这将使更多企业,无论规模大小,都能受益于AI技术,从而优化其供应链管理。
展望未来,我认为企业需要高度重视数据质量、数据安全和数据主权。他们必须确保其数据处于可用于AI分析的状态,并采取必要的措施来保护数据的安全性和隐私性。 这将是企业在生成式AI时代成功驾驭供应链的关键。
This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life.
On the show a lot, we talk about how fast generative AI is moving, right? It's like every day, every hour, sometimes almost every minute, it seems like something is changing when it comes to generative AI, large language models, and their impact on business. So
I mean, today's topic is something that moves even faster, quite literally. So today we're going to be talking about how transforming global supply chains with AI, what's happening right now and what's next. You know, this is something I don't know a ton about. And even if this is something that you think doesn't impact you, it definitely does, right?
The products we use, the services, right? When we go out to a local business, they're buying things, right? Everything is actually driven by the global supply chain. And although artificial intelligence and machine learning has a big place in the global supply chain scene, large language models has actually changed things.
that exponentially. So I'm excited to talk about that today on Everyday AI. So what's going on, y'all? My name is Jordan Wilson. I'm the host. And this thing, it's for you. Everyday AI, it is your daily live stream podcast and free daily newsletter, helping us all not just keep up, but how we can actually get ahead by knowing about everything that generative AI is impacting. So that sounds like what you're trying to do. Please
Please go to our website at youreverydayai.com. Sign up for the free daily newsletter. We're going to be recapping today's conversation as we do every day, as well as keeping you up to date with all of the AI news, fresh finds, and everything else that you need to be the smartest person in AI at your company. All right. So if you're looking for the daily news that we normally go over right before a show, it's technically a prerecorded show that we're debuting live. So make sure to check out the newsletter for that.
All right, enough chit-chat, y'all. I'm excited to talk supply chain and AI and what's happening and maybe even what's next. All right, so live stream audience, please help me welcome to the show today's guest. We have Julian Harris, the CEO and founder of RoboBuy. Julian, thank you so much for joining the Everyday AI Show. Hi, Jordan. Thanks very much. Pleasure to be here.
All right. I'm excited. All the way from Sydney, Australia. Yeah, there we go. There we go. I was going to let people know if they couldn't pick up on the down under accent. So, you know, yeah, definitely talking, taking a global viewpoint today, right? But let's start at the top. Like, Julian, like what is RoboBuy?
So RoboBuy is a supply chain analytics platform, global supply chain analytics platform. And at the heart of what we do, we extract data for large corporates. It's all B2B spend. We extract data from
All and every ERP system that they have. So one of our clients has 43 different ERP systems in their business. They've acquired this over the years from various M&A deals. So they've got 43 ERP systems and they have no single view of their data.
So the problem we're solving is they probably don't know their spend to the nearest billion dollars, and they probably don't know the number of suppliers to the nearest thousand, because we deal with very large companies like Coca-Cola in Atlanta, MasterCard in New York, really big companies with really big global supply chains.
So we extract data using APIs from all of these ERP systems, and we've built proprietary AI algorithms to clean and classify that data to a standard global taxonomy, and then surface that information in various insights and dashboards to various people within the company.
So the typical people we work with are the chief procurement officer can now see where they can make savings. They can see which of their suppliers are on contract, which of their suppliers have got purchase orders or not, how many suppliers they've got in a certain category. So if they've got a thousand suppliers in their laptop category, it probably makes sense to
pair those down to about two or three and in which case they'll save a lot of money from efficiency. So we do a lot of work with chief procurement officers, we do a lot of treasurers where we help them look at the most efficient way to pay for goods. So possibly using virtual credit cards for small payments and we work very closely with Mastercard and the banks in that area.
And then the third area, which has become bigger and bigger, is all around sort of risk compliance ESG. So we can look through your supply chain and say, hey, you've got some people that are looking like they're flagged on sanctions. Or there's people here that look like they've got modern slavery risk, possibly due to where the goods are coming from or a load of risk factors. There's some suppliers here that look like they've got cyber security flags against them.
We can tell people how much they're spending with minority groups, like in the US, a lot of people want to know how much I'm spending with veterinary-led businesses, women-led businesses, things like that, basically. So a whole suite of insights. And as usual, the big issue is, first of all, you have to source and then clean up that data because quite often the data in these ERPs is quite dirty over the years.
and then use AI to sort of come up with some smart insights on the data. So that's in a nutshell is what we do. - Yeah, no one likes dirty data, right? That's the worst part of your AI journey. But, you know, so Julian, so you founded the company, right? Was it about eight years ago? Is that right?
Yeah, 2017 we founded the business in Sydney, but we are a global operation. A lot of our clients are in the US. So, you know, I'm curious, kind of starting an analytics company in AI and like right before the degenerative AI boom of 2022.
How would you say has generative AI changed what you do, but also the bigger picture, how has it changed the global supply chain? I know that's a huge question, right? Because it's probably a thousand different ways that you can answer it. But from your perspective, what has it been like kind of the before and then the kind of after or during?
Well, that's a very interesting question, Jordan. This is my second AI company. I set up one in 2016 that we sold to Cognizant in 2020 out of London. And that one specifically focused on, it was a data science high-end AI consultancy working with the Amazon tool set. And in 2016, even though, as you said earlier, AI algorithms have been there since the 50s, a lot of the maths,
That was quite new in 2016. So we were doing a lot of really interesting work. You know, we were predicting client churn for companies. We were predicting, um,
the cost of energy for energy traders. We were doing some really, what I would say, cool stuff with very traditional AI and algorithms we've written ourselves. Same with this business. All the IP we've built is really good at doing its job, which is classifying and cleaning, spend and supply chain data.
Last year when ChatGPT comes up, it just throws the whole industry up in the air. I mean, the first thing for us is, I think, very much confusion amongst our clients. So, you know, we would say, hey, we've got very proprietary AI that cleans up your supply chain. And obviously, most executives that have been on a plane and read a magazine are now thinking that ChatGPT will do it out of the box.
ChatGPT won't do that out of the box. And it's a very different approach. So I think there's been a little bit of education there, but absolutely the pace of change now with the generative stuff is amazing. And it is just sweeping through every type of organization and every sector. And I think the automation we're gonna see for the next couple of years are gonna be quite stunning to be honest. - Yeah.
When it comes to large language models, and obviously the big step there is the ability to work with unstructured data, right? You talked about the need to not have dirty data, right? But data is data. Unstructured information is a little hard to work with, except for large language models help that. So what potential problems or opportunities are we in the
middle of matching up generative AI with opportunities when it comes to supply chain, right? Because like I said, I don't know the supply chain very well. You do. What are those areas just ripe for potential disruption with large language models? Well, people that there are a few problems you always get with supply chain. One of the first things is if you're dealing with a large buyer and their 10,000 suppliers, let's say,
The problem you often have is you don't know what your supplier's supplier's supplier's supplier is. How big is that supply chain and where does it go? So as an example, you'll be buying batteries in Chicago. Batteries have got cobalt in them. Two-thirds of the world's cobalt is mined in the Congo, probably using child slavery.
You don't know that because you bought your batteries from the corner shop in Chicago. So seeing through that supply chain is a tricky deal. We work with one company that actually has cracked that. They've got some incredibly good data and some incredibly good math. And basically they're using large language models to actually look through any layer of supply chain. So I think
You'll see solutions like that coming up. I think you'll see a lot more predictive analytics, a lot more scenario type planning. A lot of the tools today are telling you what you've done in the past and perhaps how you can change things in the future because the past is a predictor of the future. But I think there'll be a lot more tools now doing a lot more scenario type planning.
In terms of generative AI generally, I mean, once we get into the point, and people are talking about it a lot now, where AI will be writing its own AI, and I think Zuckerberg recently said that AI will be replacing his base-level coders in Meta, which is
really scary. I mean, I've been in the IT industry forever. And if you think about over the last 50 years, we've spent our time offshoring blue collar workers to China. And we, you know, we've lost a lot of that industrial base, that's America, UK and Australia. Generative AI is going to now replace the white collar workers, not the blue collar workers. So if you now take out
level of coders and all coders become you know AI prompt engineers or whatever but there'll be less of them that's going to be massively disruptive to the to the world but to start with it makes technology so much cheaper because currently it's an expensive item for most businesses but you'll end up with a lot of base coders today looking for to upscale I think you
Yeah. You know, speaking of Mark Zuckerberg and Meta, you know, a pretty, pretty recent story showed that, you know, both Meta and Apple are trying to enter the, you know, AI humanoid robotics space, right? A space that, you know, there's pretty promising, you know, prototypes out there from, you know, Tesla Optimus, from Figure, all of these other companies. Yeah.
You know, at least here in the US, it's I mean, this is a space that's hot on fire. What are your takes on on that? Right. Like, especially as it comes to the supply chain to logistics. Right. Is this something is it our factories and warehouses going to be just humanoid robots in the near future?
Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on Gen AI. Hey, this is Jordan Wilson, host of this very podcast.
Companies like Adobe, Microsoft, and NVIDIA have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead. And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Gen AI. So whether you're looking for chat GPT training for thousands,
or just need help building your front-end AI strategy, you can partner with us too, just like some of the biggest companies in the world do. Go to youreverydayai.com slash partner to get in contact with our team, or you can just click on the partner section of our website. We'll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on Gen AI.
Well, I think if you went and saw an Amazon warehouse, you'd see that it's pretty automated now. I don't know if you've seen videos of those robots. I mean, for sure, the physical supply chain is getting more and more automated by robots. I mean, that's a given. I think what's going to happen is, as I said, there's a lot of what I call quite
traditional software that's out there. And this is in every space. This will be Salesforce, let's say, dominates CRM. SAP dominates the ERP industry. I mean, in the world of supply chain, at the top end corporate, they're either using Oracle or they're using SAP.
Those guys have got nearly 50-50 world share. There's some other things like Microsoft Dynamics and a few other tools there. But at the top end, they're either on SAP or Oracle.
And those tools have been around for probably 40 or 50 years, I would have thought, certainly 30 years. Now, those guys are scrabbling like crazy and you'll see all the ads and the PR out there to transform their products into AI products. But obviously, you're effectively adding onto a code base that's still very old. So I think over the next few years, there are going to be some pure play AI players
powered by generative AI that just say, look at the market share that Salesforce has got in CRM, or look at the market share SAP has got in the ERP space, and quite frankly, probably take them out. I think they've got a very, very strong, a big stronghold at the moment. But I think generative AI, particularly when you get to the point where AI is writing itself, and it's highly automated,
the price of producing software becomes like next to nothing. - Yeah, you know, and I'm curious, you know, you work with a lot of big brands that have a huge global supply chain footprint. What are some of the most common
you know, not mistakes, right? Because that's, you know, dirty data happens, but what are some of the most common, you know, redundancies or some of the most common shortcomings that, you know, usually have a generative AI solution? Like, what are these things that you're seeing over and over? Because maybe that can help us understand, you know, where this intersection is ultimately headed. Sure. Well,
Supply chain, I would say, is quite traditional, as in it's probably one of the last areas to change. There's a lot of
There's still a lot of manual input into these ERPs. You know, people getting in. This is where the dirty data comes from. Somebody in the accounts payable team in a large corporate will get these invoices coming through. They may be scanned. Hopefully they may not be. They're double entering into SAP. They make mistakes or they're really busy. They've got to enter 500 a day so they don't bother putting in the details. You know, instead of
saying what each line item is that they bought, they'll just put stuff at the top, bought some stuff, 10,000 bucks. Now, when you're then trying to classify that data for some execs to look at what they're paying in each category, it's very hard if somebody has just titled something stuff.
Yeah. So, you know, you need to go back and look at the supplier, perhaps go and scrape some data from the Internet about the supplier, you know, throw that into a model and start start predicting things that way. But, yeah, I mean, the industry has always been, you know, quite traditional, quite slow at adapting to some of the technologies and not as joined up as it should be.
For a while, we've had AI that can read contracts and read documents and things like that.
Certainly large language models are making that way more accurate and way more efficient. So, you know, I think to automate the end-to-end supply chain without these gaps where humans are double entering things would make a massive difference. But I think somebody, yeah, certainly using large language models, it's right for somebody to take just a totally different approach to
to the supply chain yeah so you you know we've kind of touched on or at least the way i see it you know on some of the front end of the supply chain so you know manufacturing warehousing what about on the on the latter side right so when we look at uh transportation distribution uh customer delivery all of those things where do you think we're headed next right like i i've always thought like you know 20 years ago if i would have thought about this i'd be
be like, oh, we're going to have fully self-driving cars and drone deliveries and all these things. It doesn't look like we're quite there, right? But at least when it comes to large language models, how are they going to find their place in the near future in those other places, such as transportation, distribution, customer delivery, et cetera?
Well, I think there's a lot of work being done in that area by the real specialists. I mean, Amazon is probably the leader, I'd say, in those sort of logistics. And I think individual companies just can't spend the way that somebody like an Amazon can in that sort of technology. So you see a lot of the more sophisticated companies outsourcing a lot of that logistics to people like Amazon. And they are trialing drone deliveries, you know,
you will get automated vehicles delivering this stuff on show. I mean, the drone deliveries, I think that Amazon are doing, the trials have been quite successful. When it moves to mass rollout, I don't know. But I mean, all of those are now controlled because obviously there's a navigation piece, there's a safety piece. All of these are now controlled and going to be controlled by large language models for sure.
So, you know, one thing that, you know, I'm always looking at is how generative AI specifically can help even non-technical people make better decisions.
How does that ultimately play out in your space? You know, when it comes to predictive insights, scenario planning, right? Is this something that maybe it's only for the big global players when it comes to using this in their supply chains? Or are some of the smaller guys able to finally get some of this, you know, technology because of large language models that has maybe only been afforded to those companies that could, that had big data teams for decades?
No, I think the technology, you know, starting with chat GPT and now, you know, all the others that are coming fast behind, it's becoming very accessible, very cheaply for just your average worker. I mean, everyone now that's, you know, got a Microsoft laptop has got, you know,
co-pilot available to them, you know, and co-pilot will just, I mean, they drop in new versions every other month as it appears. I mean, that will just get more and more powerful. And it's, you know, it's designed for the business user to use, not a techie, basically. You know, Anthropic did a big deal with Amazon. So Amazon, you know, pushing more and more products out there. I think, I mean,
everyone's talking about this year we're moving from chats chat bots effectively AI chat bots to AI agents which essentially are just smarter versions but essentially these are going to be in the hands of the user I mean
Just from an efficiency point of view in my company now, very few people are building PowerPoint decks. They're telling AI to build a PowerPoint deck. Very few people build the start of a Excel spreadsheet. They get ChatGPT or Copilot to build it and perhaps just do a little bit of editing. I mean, I think most people, like just office workers, are getting into the groove and people like Microsoft are going hard at
free training for all these people. You know, it will take a while, a little bit of a culture shift, but certainly, you know, the next generation of workers coming in, if there's one skill they need, it'll be how to how to maximize the use of AI. Otherwise, they'll get left behind.
Speaking of what's next, I have to ask you that. So as someone that has two separate AI companies, you said your other one was acquired and you have a successful company now. What do you think is next? Or what is maybe keeping you up at night with whether it's excitement or worry, at least when it comes to generative AI in the supply chain? What are you looking at?
So, okay, so I'll start with my or our suite of products. It is excitement and worry, as you say, because we built some very specific models over the last eight years. And obviously the question in your head, as soon as ChatGPT pops up is, is that all now replaceable by this simple LLM that's popped up?
and rendered our whole product worthless. So, you know, I think everyone at the moment are doing a bit of navel gazing, including all the big players and looking
stripped down all their product and understanding, okay, what have I got that's defensible and what actually could I replace with a large language model? Because obviously, if there are parts of my suite of products I could replace with a large language model, and that's what we're looking at now, in many ways that makes it more efficient and easier to maintain and has more features around predictability
you know predictive and things like that basically so i think most people are looking at that i think as we said earlier i think some of the big players must be very worried if i was sap and oracle i would be worried that somebody's just going to come in and build a pure ai version of the erp because that is you know a massive multi-multi-billion dollar um business um
there's obviously going to be a ton more automation so in terms of supply chain you know typically
if you're in a coca-cola or any or a walmart they'll you know they might have hundreds of category managers analyzing each category under a procurement um head um a lot of those jobs will disappear because a lot of the that stuff will just be automated using large language models you know what to buy who to buy from which suppliers to let go which suppliers to spend more with
all that is easily automatable with this sort of product. And I think, as we said earlier, I think you'll see a lot more predictive analytics coming out with the
with the large language models and scenario planning and things like that. A lot more intelligent, forward-facing stuff. All right. So, Julian, we've covered a lot in today's conversation, but as we wrap up, what do you think is the one most important thing? And this is a big question, so answer it how you may. What is the one most important thing you think people need to keep in mind when thinking about the supply chain situation
generative AI in where we're headed next? The single biggest problem they're going to have is where is my data? What state is it in? How do I get it into a place where it's even usable for AI? Because it absolutely, I agree with you, large language models can deal with unstructured data, absolutely. But if that unstructured data is fundamentally rubbish,
it's still going to have problems. Yeah. I mean, we've been, you know, we've built search engines and all sorts of things in the past. Unstructured isn't necessarily a problem. There's technology that's been doing that for a while, but the quality of the data. So with most people, they've had hundreds of different systems. They're not joined up. They need. So the first job is where's my data? How do I access it? How do I clean it up? Once I've cleaned it up, how do I make sure that I've got policies in place that it is cleaned up?
So it's all around data. I mean, the other thing to think about is data sovereignty and a bunch of other things around that. There's too many people at the moment will throw their data into something like chat GBT, unaware that they're now sharing that with the world.
Now, you can still use the models on your own infrastructure or private clouds and there's ways around that, but people really need to think through their data from where is it, how do I access it, how do I clean it up, how do I make sure I've got policies to keep it clean and accurate going forward, but also how do I secure it and make sure that all my privacy settings and security settings are right.
All right. So a lot to think about there. I think some great takeaways, whether you're in the supply chain or not. I think today's conversation is an important one to have. So Julian, thank you so much for joining the Everyday AI Show. We really appreciate your time.
Thank you, Jordan. My pleasure. All right. And as a reminder, y'all, we covered a lot. If this was helpful, if you're listening on the podcast, please go ahead, click that subscribe button and leave us a rating if you could. If you're listening online, share this with someone that's in the industry. Keep them up to date.
Let them know what's happening and what's next. Also, if you haven't already, please go to youreverydayai.com. Sign up for that free daily newsletter. We're going to be recapping today's conversation with Julian and a whole lot more. So thank you for tuning in. Hope to see you back tomorrow and every day for more Everyday AI. Thanks, y'all.
And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit youreverydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.