Factory Intelligence: Lauren Dunford on AI, Margin, and the Manufacturing Fundamentals

Lauren Dunford:

What we see is manufacturers successfully scaling are focusing it on using technology in the right way to drive margin by attacking the fundamentals. We use the NASCAR analogy of, in a NASCAR race, it's not going to be the shiny object, shiny aerodynamic wing. It's going to be the driver, the tires, the pit crew execution, all of that unsexy stuff. We see people focusing on getting the fundamentals right and leveraging the technology that drives that in the smartest way and focusing on working backwards from the business.

Jeff Kavanaugh:

I'm Jeff Kavanaugh with the Infosys Knowledge Institute, here at the World Economic Forum's Advanced Manufacturing Supply Chain Forum in San Francisco. Joined today by

Lauren Dunford:

Lauren Dunford, CEO and co-founder of Guidewheel. And so excited to be here.

Jeff Kavanaugh:

Great to have you. Do you own the models? Do you own data? Do you own algorithms? Or what is it that people are buying from you?

Lauren Dunford:

So we own the models and they're getting smarter with every machine. The customers own the data. So the data coming from their machines, they own. The analytics are better on the data, so we can make the data much more useful across the board. I'll give examples in a moment. The other thing that the AI does is it makes it much faster and easier to connect equipment. So I was on the phone this morning with somebody who has 1940s hammer presses from World War II. In the past, those would have been impossible to connect into your ERP, your other systems on the plant floor. What we've got is an algorithm called State Sense that is absolutely best in the world at taking the power going into any piece of equipment, including very old equipment, and translating that into what is going on with that equipment that matters to the business.

Jeff Kavanaugh:

Would you say that your market then is more of the small and medium enterprises?

Lauren Dunford:

So it started with small and medium enterprises because two years ago, before AI got as good as it is, what we were selling was just visibility. You've got your toasters, suddenly you can see what all the toasters are doing and why. Now, because of the ability to take that information, combine it with the business data in an SAP or ERP, and be able to not just give data but give guidance. You're a plant manager waking up in the morning, here are the top things you should pay attention to because they're the bottleneck that's most valuable.

Jeff Kavanaugh:

Visibility to what happened. Guidance on what to do about it. And then maybe the next shoe to drop will be actually doing it, the agent.

Lauren Dunford:

Closing the loop. We see it as visibility, prediction. That predictive is nice, but really people want prescriptive, what do I do about it? What worked, what didn't, what was the impact? That's the closed loop. Whereas we get that closed loop, it tees up exactly what you talked about, which is right now this is humans and machines on the plant floor. Over time, what most businesses are going to do is look backwards from how they deliver value to their customer and think about the different skills and things that need to happen to do that.

Jeff Kavanaugh:

So then it becomes an input into strategy, planning, and maybe your operating model tuning.

Lauren Dunford:

Operating model and resourcing, because when you...

Jeff Kavanaugh:

Part of the operating model, one chunk of it's talent. So that just kind of was cascading down.

Lauren Dunford:

And talent being increasingly leaders thinking about not just human talent, but what's an AI only, what's a human plus AI.

Jeff Kavanaugh:

Capabilities needed and scale.

Lauren Dunford:

Exactly. And so this is laying the groundwork in an AI-ready way, not assuming that people are doing that aggressively today. Not everybody is on every plant floor in America. Let's drive some value right away by getting the right guidance. Let's fill the hole for people that don't even have visibility yet, and let's do it in a way where how do we make it so that it can be learning in a safe and anonymized way to do the things everyone should do, like compressed air leaks, let's fix the compressed air leaks, let's make the CapEx investment where modernization is decarbonization, let's drive the productivity improvements so that we can actually have smaller plants, small and medium manufacturers, as part of the supply chain of those much larger manufacturers, be able to be keeping up and making the resilient, awesome industrial ecosystem that we need.

Jeff Kavanaugh:

What separates manufacturers who are successfully scaling AI from those stuck in pilots?

Lauren Dunford:

What we see is manufacturers successfully scaling are focusing on using technology in the right way to drive margin by attacking the fundamentals. We use the NASCAR analogy, in a NASCAR race, it's not going to be the shiny object, shiny aerodynamic wing. It's going to be the driver, the tires, the pit crew execution, all of that unsexy stuff. We see people focusing on getting the fundamentals right and leveraging the technology that drives that in the smartest way and focusing on working backwards from the business.

Jeff Kavanaugh:

Last time I checked, especially the medium and small manufacturers, don't have just a bucket of money saying, oh, this is just in case we need to grow and expand. They don't have it. Where do they find the funds to do these things?

Lauren Dunford:

For us, we have to come in with a wedge that drives immediate impact. What we found is the technology's a part of that. Humans, making sure it's deployed successfully and used, are another part of that. So we have a whole value realization team that when we're signing up to work with a customer, we are holding ourselves to this, this has to not only pay for itself but drive a great...

Jeff Kavanaugh:

So you've got pretty tight models about the value creation. You have to.

Lauren Dunford:

Oh yeah, we have to. We've just seen time and time again technology thrown over the fence, it might work. Someone grabs and runs with it. Our customers are getting woken up at 2 a.m. from a quality issue, having to go into the plant, pull everyone to figure things out.

Jeff Kavanaugh:

That's assuming that people will show up to work when they are supposed to.

Lauren Dunford:

Our customers are juggling so many things. So our job is how do we make pieces of that easier and easier while driving the business value because there aren't extra funds sitting around.

Jeff Kavanaugh:

How are they measuring the ROI?

Lauren Dunford:

We're starting with very simple things. So we say, okay, is it increasing the dollars coming in? Let's go back to that toast factory. Are more dollars coming in or are fewer dollars going out? Who's going to make that change that actually moves the needle in the P&L? And what do those people need to make that change? So if it's driving throughput from a capacity constrained line that then opens up the ability to work through a backlog and drive revenue, fantastic. We love that situation. If it's not a capacity constrained situation, who's going to make what changes based on driving improved productivity? And if there aren't changes that will be made based on driving improved productivity, let's wait until there will be, because it has to pay for itself.

Jeff Kavanaugh:

What role will AI play in building more resilient and adaptive supply chains?

Lauren Dunford:

Resilient and adaptive supply chains require a bunch of small and medium manufacturers. You can't ship a plane without the right bolts. You can't get a car on the road without the pieces that make it happen. And so one of the ways in which AI is contributing already is allowing people to be able to get value so much faster and easier. So that State Sense algorithm being able to clip around the power and make sense of that, like a Fitbit or smartwatch, of getting to the right guidance without tunneling into the machine PLC, opening up attack surfaces all over the place, and costing millions within an 18-month timeframe, that's already allowing there to be visibility and productivity in areas of the supply chain that would have been hard to reach.

Jeff Kavanaugh:

We got to talk about cybersecurity. What are you seeing there and how, because these tier two and tier three can't afford and don't know these things, and yet that's probably where the hackers can come in.

Lauren Dunford:

The hackers go straight for manufacturing. It's real, I mean, I think it's one of the top sectors they go after. So from a cybersecurity perspective, what we're seeing is people caring a lot more, which is good. We've always been trying to work backwards from how do we design the way this works in a way that's secure by design. We always start air gapped, for example. Don't touch any pieces of the equipment. Don't even touch the network. Let's not open up a new attack surface. Let's keep this separate. Let's drive impact in a really safe way. That has been very nice in this time where everyone cares a lot more. We also right now really think about, let's be very careful about starting with human in the loop. Let's not let agents loose where we don't yet know the things that might happen. Let's be very careful about those deployments. Think about what's going to be the right answer. And this is kind of working backwards from what our customers need. Sometimes it's not an LLM getting loosed on the data. It's how do we actually just think about the smart deterministic algorithms that make sense?

Jeff Kavanaugh:

If you go back just a few years, even machine learning is kind of a middle ground. Because it's not deterministic but it's not the wild west of these LLMs either.

Lauren Dunford:

We've got this nice Scout algorithm where it learns, when there are 15 different ensemble models running behind that heartbeat data from the equipment, it'll say, hey, this is actually big enough to care about, send an alert. The customer says yes, that did or didn't move the needle for me, it'll learn based on that. We also have, and that's just classic machine learning. We also have this throughput forecasting algorithm, which is also still very classic machine learning, but comes across as much more advanced because it's actually taking the production data of what happened in this shift from the ERP, learning based on that boring electrical heartbeat of the equipment, to translate the electrical heartbeat into, as we run through the shift, what's the most likely best case, worst case, landing at the end of the shift, end of the week, end of the month. And being able to then take that and improve each time, these are all things where it doesn't require the LLM approach.

Jeff Kavanaugh:

We've got to talk about humans and AI. What do you see about the changing relationship on the manufacturing platform and with this, the managers, let's say, as AI and agents play a bigger role?

Lauren Dunford:

Right now it's still on most plant floors not a combination of humans and agents making decisions. It's not like AI has diffused onto the plant floor in a crazy way. So right now, we're not seeing dramatic changes. It's better analytical tools. Like we just released this new Wheelie where a plant manager can get automatically, hey, what are the things I should focus on today? Surface that, get that right kind of guidance. That's what we're seeing today, being the types of use cases that are most prevalent. We're not yet seeing humans plus agents, you know, an agent based on that throughput forecasting saying, hey, bring in some more material. Based on the patterns and how the lines interact, where the bottlenecks are, that's where it starts to get really interesting because there's so much complexity in a lot of these organizations.

Jeff Kavanaugh:

All right, since we are here at the World Economic Forum, I've got to ask you about the Lighthouse. How do you see what the World Economic Forum is doing, helping especially the small and medium manufacturers, but in general just providing this tool to level the playing field?

Lauren Dunford:

Oh my gosh, I mean it's so important. So what the World Economic Forum is doing to pull best practices and make it concrete, no one wants to hear improve energy efficiency or improve productivity. No, how? What are the situations where it creates the ripe environments for certain types of technologies or certain types of, it might not even be technology? It might just be process improvements. And being able to make that accessible to everyone, small and medium manufacturers, there's so much, and what I just love about manufacturing is there's so much that's different. It's all cool and interesting and so different plant to plant or company to company, there's also a lot of fundamentals that are the same. And so if we can work with the fundamentals, every factory pretty much has compressors and chillers and this type of fundamental supporting equipment. Every factory also cares about the goal and constraints and bottlenecks. How do we take those things that are fundamental, create really concrete, valuable, here's what to do and what the result is, and make that accessible to everyone in a way that's digestible.

Jeff Kavanaugh:

Well, besides being efficient, it also means what used to be the repetitive, so-called mindless wrench turning becomes almost a utility and then the fun, interesting things, which actually are more interesting than other industries. And in manufacturing being the noble, interesting, cool profession emerges.

Lauren Dunford:

Yes, exactly. So then you can have talent coming in and I don't know what the stat is about the percent of people that have gaming skills coming out of school today. I think it's like 55%.

Jeff Kavanaugh:

Oh, you do simulation, you are basically video gaming.

Lauren Dunford:

Exactly. So if you can come in and you can say, oh my gosh, I'm going to have a truck or a plane or chocolate candy coming off my line at the end of the day, and I'm going to take all my different constraints and run the simulations and course correct in real time, suddenly that has immense potential to be fun and to be...

Jeff Kavanaugh:

Oh, it's fun and it has a physical part of it where people view the cube as a prison cell. I think it's a moment for manufacturing.

Lauren Dunford:

I think it's a moment for manufacturing. For any parents out there, for any Gen Alpha or whatever it is of people entering the workforce, come not only take the great technology and have a lot of fun building real things with it, but come help us build the right technology or create your own company to build technology. So if people are looking for careers that are full of potential moving into this new world, to be growing and more interesting by the moment, manufacturing is such an exciting place.

Jeff Kavanaugh:

Thanks, Lauren, for a great discussion. I'm Jeff Kavanaugh. Keep learning and keep sharing.