From Copilots to Agentic AI in Manufacturing — with Lina Huertas

Trend Detection Podcast ·

Show notes

Welcome to the Trend Detection podcast, brought to you by Senseye Predictive Maintenance – which gives you visibility and insights into all your assets, from single machines to full plants to help you reduce downtime, increase knowledge sharing and accelerate digital transformation across your organization.

In this episode, we’re joined by Lina Huertas, Industry Executive for Manufacturing at Microsoft UK, to explore how generative AI, copilots, and agentic AI are reshaping digital manufacturing — not just speeding up tasks, but changing how work is designed, delivered, and governed.

We unpack the difference between copilots (which assist and enhance human work) and AI agents (which can complete tasks end‑to‑end within defined boundaries), and what this shift could mean across the shop floor, engineering, and back office.

You’ll learn:

How copilots and agentic AI differ — and why that matters for manufacturing workflows and roles.

How organisations are thinking about moving from assistance to more end‑to‑end task execution (with human oversight and clear boundaries).

Why human–AI collaboration is becoming a core capability, with work shifting toward supervision, decision‑making, leadership, and critical thinking.

The key barriers to scaling AI in manufacturing: data silos, fragmented systems, shadow IT, and organisational structure.

The skills manufacturers (and individuals) need next: hands‑on AI literacy, “learning how to learn,” and leading in a workforce that increasingly includes AI systems.

You can find out more about how Senseye Predictive Maintenance can reduce unplanned downtime and contribute towards improved sustainability within your manufacturing plants, by visiting: www.siemens.com/senseye-predictive-maintenance

Connect with Lina on LinkedIn:

https://www.linkedin.com/in/linaahuertas/

Transcript

Manufacturing, it all comes down to engineering. And when you think about engineering, yes, you design the product, but actually more and more, the process that you use to do your, your product and the material that you use are, um, innovation factors, and they can completely change everything from then on. So innovation in materials is another really interesting topic in terms of technologies. Welcome to the trend detection podcast powered by Siemens, and now your host, Nel Sullivan. Hello and welcome to the latest episode of the Trend section podcast. Um, I'm really pleased to have a very special guest. All of our guests are very special, of course, but, um, this, uh, Lena in particular is a very special guest because we were just talking before and it's been at least a couple of years since we've spoken on this podcast, your second time on, but there's been a very big gap, but it's very nice to have you back, Lena. Welcome back. Well, thank you for having me again. I feel privileged. Oh, excellent. And so Lena is, um, is from, um, from Microsoft, and I'm gonna let you sort of introduce yourself to the audience in a second, um, but we're here, sort of, I guess revisit our first conversation. I think I'll probably link to it in the, um, in the show notes, uh, uh, you know, when we publish this episode. So for you to sort of get the context of it, we're going to revisit some of the themes, um, that we discussed then, and we're gonna talk about the future of digital manufacturing, if we can really do that, even with 6 months in advance, 1 year in advance, it's very difficult, but we're gonna try and sort of dive into that, cover some of the topics we covered before and see where we are today and, and go from there. So, um, but I guess before we go into that, Lena, it'd be great to get a bit of an introduction from your side and your role at Microsoft. Uh, thank you, Nia. Thank you for having me. Um, yeah, I'm Lina Huertes, and I'm a, um, as I said, I, uh, you can pronounce my name Fuertes or Huertas. I'm an industry advisor at Microsoft and, uh, what that means is I am part of the industry team. We have teams that are vertical and specialize on the specific verticals because we recognize the importance of, um, Understanding in depth the outcomes that customers are trying to achieve and in this case manufacturers and obviously that varies greatly from Uh, vertical to vertical, so we focus very much on business outcomes and those processes that are more directly, um, linked to those outcomes. So focus mainly on usually engineering, design, engineering, manufacturing, so the factory production side of things and supply chain as well as connected product, uh, increasingly, which I know one of your questions might be around that. Exactly, exactly. We'll certainly, we'll certainly come to that as well. But I mean, to, I guess, open up the conversation, I think at the heart of our last conversation was about, you know, the, the technologies that are transforming manufacturing. It was very interesting because I think we were, when we last spoke, generative AI. And, and that sort of things was just sort of starting to come into everyone's vocabulary and as we know now, it's sort of everywhere. So, I mean, we could maybe talk about where we are today with generative AI co-pilots, maybe even agentic AI, um, and any other sort of technologies you see at the moment leading that are starting to sort of creep into, I don't know, into manufacturers' minds and headspace at the moment. Yeah, so, I mean, since we last spoke, even though generative AI was, was, I mean, generative AI has been around for a long time, but I suppose it got to people's fingertips with, uh, transformational models made available to through a consumer interface, which had GBT and that had just been released. Um, but I think what we've seen in the last. 18 months or so has uh really transformed what organizations are doing and how we're interacting with them because, um, this whole topic of agentic AI is um is quite different to previous technologies to see what we saw usually technology, you know, is a. Tool and we say, well, the important thing is what you're going to do and try to achieve rather than the technology. The technology is an enabler, and you can, you have a process, you get a technology and you can do it quicker or more accurately or better in a way. And I think the big difference now with the Gtic AI is that is still an enabler, but it's changing things so much in terms of how work is delivered or how work could be delivered that is given the organizations the opportunity to actually rethink, you know, what are the capabilities that we're trying to deliver to the market. And how do we rethink how we deliver them, almost from scratch, like if, if you think about whole functions like procurement, you know, if, if we have no one in the function, and we were. And we were considering Agente AR as the default mechanism for delivery. What would the function look like, and it probably would look like pretty different. So this is not just helping people do the same job quicker or or or or just accelerating or improving a process. It's. transforming processes and functions and organizational structures and leadership styles. So it's a, it's been quite, um, disruptive, and that, that, that's the probably single biggest things we see changing. There are other things on the fringes that we see are going to be enabled by this. And so, for example, one of the things that I always think in the space of manufacturing and engineering that has always been. Super interesting, even though it hasn't been my topic particularly, but um materials is something when you think about. Manufacturing, it all comes down to engineering, and when you think about engineering, yes, you design the product, but actually more and more, the process that you use to do your, your product and the material that you use are, um, innovation factors, and they can completely change everything from then on. So innovation in materials is another really interesting topic in terms of technologies, but when you look at the combination between. Material innovation and the possibilities of materials and the way you can innovate with agentic AI and quantum maths and potentially in the future quantum computing, I, I, I see that, um, manufacturing and products can look, um, quite different and can be transformed, um. Uh, it's an order of magnitude that that probably quite hard to predict at the moment, because, you know, you're, you're thinking before, you know, new materials, new alloys, new, new chemical compounds takes years to come up with a new formulation, whilst now it can take you months, weeks, or even hours, depending on the case with all of this too. I, I think that's really going to change what we can do. And I guess maybe take a step back a bit. I'm, I'm sometimes guilty of maybe in my own head, presuming that whoever's listening out there knows Agenic AI, you know, and, and maybe they, they do, maybe they've seen it on social media or seen it in the news or seen it wherever, but maybe we could start by sort of defining that a little bit more, um, Lena. And also how it's at the moment, maybe it's still in its infancy, maybe, but how is it redefining sort of that, you know, people in their day to day role on the shop floor, for example, just examples you're hearing. Yeah, so, so agents that there are different types of agents. Basically agents are. Um, digital entities, they can help you deliver part of a process or an entirely, entirety of a, of a process. And, and as I said, there are different ways to do it and we see that reflected in how organizations are transforming and usually the first step and the first type of agent would be a retrieval agent. That is basically helping people the same job they were doing before, but probably doing less of the data intensive tasks, accelerating the job, helping them basically increase personal productivity. Um, so there you can find from, from our site, Microsoft 365 compillo, and we were discussing it earlier, and you, you and I are both avid users. of it. And as we saw in the call itself, you know, you have a question you can't remember, you ask, when was the last time I met Nile, and when was the last podcast? What we, what did we discuss that, what, what was this quote, and, you know, summarize this thing, how we prepared for the podcast, and there are other more interesting use cases as well. So it's that, that's the first, so it's an agent that is able to leverage. Information and reasoning capabilities to help you give you an answer or perform simple tasks for you like summarizing things or preparing for things. Um, but the, the next phase, which is where it starts getting a bit more interesting, is when you start letting whole letting agents deliver whole processes for you, so you can think about, um, some simple processes, for example, Invoice, uh, invoicing customers. And there, there, there are companies that are already looking at, well, there is a simple process. If you look at our system, we already know what are the triggers to invoice a customer and what are the um the data fields. So let's just let it release and if there are any exceptions, let's raise those exceptions and, and you can then see that when you do that, the job that people deliver. Really changes because it goes from having to do this thing to it's a very, very repetitive thing. You probably need to do hundreds of potentially hundreds of invoices a day, and if you think about that, it's not a very fun job to actually just letting the agent to do that, and you manage by exception. And it's, it's interesting because, well, the third phase is when you start letting agents negotiate and interact with other agents, and you can start then automating or digitizing whole functions. Um, but what's, what's interesting is that that's when the, the job of a person changes because you go from having a mix of Some really data intensive boring kind of all the drudgery of your job that we, we, we all have to do that nobody enjoys and a mix of things that are really quite um important strategic defining a strategy, making decisions, uh, defining and negotiating trade-offs between teams, doing teamwork, kind of understanding people, how negotiations with, with, with other companies, um. And sometimes I think it's, uh, you can easily, um, interpret this as replacing people and I think there's a lot of talk about that. Um, but actually, when, when you look at what people used to do and what they have the potential to do is actually putting people at the center and really getting the most out of human talent because all of a sudden. Hopefully we won't have to do any of those tasks that we don't have to do, and we can focus where we really add value, defining work, training agents, defining what they're going to do, setting objectives, supervising, making decisions, um, and doing all of the soft, not soft is probably the completely the wrong way to referring to, but doing all of the strategic, uniquely human. Things that we can do as part of that process and then you're really optimizing human talents and transferring all the rest to AI. So sorry, that's a bit of a long-winded way of explaining agents so that that's just how we see them. No, no, no, not long-winded at all, because I think, yeah, because sometimes it can be very high level as well, you know, some of it, you know, when you talk about an agent, I think people might know it in one sentence, but actually, what's the, what's the real detail? What's actually driving at the back, and, and you've explained that very well, and there seems to be sort of different sort of stages, it, you know, you know, of implementation, let's say, um, for different companies as well. But what you, you, what you were saying, leader, actually led on quite well to what, what I was gonna ask. And what I see from what you're saying is, as well, there's, there's a certain collaboration. We talk about collaboration between partners, Siemens, Microsoft, other partners, etc. but there's almost like a human and AI collaboration, if you want to call it that. And a new skill set required for the, you know, for the human in the loop here as well. And so maybe it's in, again, I'm, it's a bit of a long-winded question as well, but what, uh, with this new collaboration in mind, what sort of new skills are required from, from the human point of view to get the, the most out of the AI and get the most, be as productive as possible, I guess. Yeah, that, that, that, that makes a lot of sense. And I think it's one of the key things as you, as, as, as we all progress through this journey, how, how do we, what do we need to learn? And I think one of the, maybe a bit philosophical, but most foundational things is kind of, um, learning to learn. And I know that's a bit of a cliche, but I think there's definitely an element of, uh, all of us. Kind of accepting that we're all gonna be um. Almost new at this again and that nobody is the real expert of any of this. The technology is moving so quickly that I think we all need to become apprentices again and almost relearn our jobs. So that I, I would say that is the first thing. The second thing is I think we all need to get. Hands on with the technology and understanding a bit more because um we're only going to be able to work with the technology if we understand it and we understand the the actual capabilities of the technologies, how it works, why does it work like that under the cover, what's possible, what's not possible. So I think there's going to be an element that's why one of the, the first things we, we say in the journey is, you know, and. Empower people with their co-pilot, with an assistant, so that they can, one, become more productive, but also by using it. I think you've probably experienced it yourself. I certainly have. Uh, there's a little bit of a curve in learning to use it and understanding, you know, what, how do you use it, and, and what is it good at and what is it not, not good at. And I think that gives you an element of awareness to then go to the next stage, you know, I'd say, um, And this is not necessarily new, you know, if you looked at the reports from the World Economic Forum and McKinsey around the skills that are going to be important, um, it's a lot of the skills and what, why the reason why I say soft is because I think traditionally they have been referred to soft skills, but we probably need a new term is the things around critical thinking, collaboration, negotiation. leadership. I mean, leadership is going to be key, and it's not just going to be leadership of people, but actually leadership of agents because they become part of the workforce, and they're like anyone else, you need to recruit an agent who decides that you need an actual agent. You need to shape the, define the scope of the agent. You need to help deliver that agent. You need to make sure that you're assessing that agent and Helping the agent improve because these are not static and that's one of the um advantages, but also I suppose the responsibility that you need to make sure that it's growing and developing in the right way, delivering the right objectives and supporting those agents. So I think leadership is going to be um another thing and, and the last thing I think it's probably going to be a different answer for everyone depending on their job it is then. I think everyone has a responsibility to question. What is my job gonna look like in the next 11 or 2 years, seeing where the technology is going, and where do I need to focus my efforts and on, on the basis of my specific job, what other skills do I need to take and I think, so we're all going to have to be a bit more proactive and being kind of driving that ourselves, because otherwise, probably the organization is kind of going to do it to us. It is always better when you do it yourself rather than have it done to you. And I, I, I guess in a way that's, that's quite an eye-opening, you know, conversation with yourself if you think about what's my role gonna be like. Not many people, uh, for many reasons, don't, you know, don't want to maybe look at that for thinking it's a, a negative thing, but actually it can be a very, it should be a positive thing. This is, you know, this is what's happening, these are the shifts in the industry, and this is where I want to be, and then try and align that together. I guess it's something like that without being an expert in that area, but, um, it's about sort of making the conversation a positive. This is what I can do with these fantastic. Techno this fantastic technology we've now got to back us up, exactly. 100% because you can do so much more and and this is, there are two things that are really important. One is, you know, The culture in organization is going to be really important to make sure that people feel safe enough to transform their own role and therefore accelerate the transformation of the whole organization, um. Yeah, that, that, that's gonna be um Really cute. There was something else that I was going to say, but it's just flew off. It will come back. That's all right. That's alright. It just proves that we're live. We're live. We're real people and the problem is we're not just avatars, and I thought it would be me that would do that, Lena, to be honest. So, um, so don't, don't, don't worry at all. Don't worry at all. But, um, but I guess, and again it brings us nicely um along to. Um, sort of, maybe some of the barriers, because all this technology and all, you know, there's some fantastic use cases and there's organizations doing amazing things, but, of course, in order to implement technology and to use, utilize it in the, in the right way for success, there's, you know, there's, there's sort of barriers or there can be supposed barriers to be overcome. Um, and I know we previously talked about things like, you know, culture. I mean, we always, we always know that cultures. And culture shifts an important part and also this idea of being stuck in, um, proof of concept purgatory, um, as well, uh, if you recall taking that, is that, are those still sort of the key aspects or is there, is there more now with the, the more, you know, with Vigentic AI now I forgotten that we discussed that, but the, they're the first things that, that came to mind is, is fragmentation from a technology perspective or from a Organizational perspective, so we still see, even I think there are quite a few reports out there that say, you know, 94 or 70% of AI projects are failing, this is not working, this is just a bubble, and I think what we see is, is actually, yes, the technology is there, the technology is reliable, but it's. It's sometimes easy to go down the path of in isolation, actually, let's go and let's, let's try it, and it might work for a specific example that is quite bound and maybe with not a lot of interactions, but the question is how do we make it scale. And scaling is the hard thing, so building one pilot is, is, you know, the technology works. It has been proving it's probably the fact, well, there are figures that say that it's the fastest adopted software category agents that. we've ever seen, so adoption is there. The technology works, it's delivering, but I think for organizations to really reap the benefit, they need to scale, and that relies on having the right foundations from a technology perspective. So things like Data silos are very common, particularly in manufacturing, as, as you, you probably know this from your, your background, you know, you have some data in the maintenance system, some data in the MES system, some data your SAP, some other data trap in the machine. is that, well, how do you reason across that? …

Full transcript (WebVTT)