Best of LinkedIn: Smart Manufacturing CW 47/ 48

00:00:00: This episode is provided by Thomas Alguyer and Frennis based on the most relevant LinkedIn posts about smart manufacturing in calendar weeks forty seven and forty eight.

00:00:09: Frennis is a B to B market research company that supports enterprises across the smart manufacturing industry.

00:00:14: with the market, customer and competitive insights they need to navigate dynamic markets and drive customer centric product development.

00:00:22: Welcome back to the deep dive.

00:00:24: Today, we are immersing ourselves in, well, the absolute forefront of industrial automation.

00:00:30: Yeah, there's a lot to unpack.

00:00:31: There is.

00:00:32: Our mission is to take that firehose of information from LinkedIn in weeks, forty-seven and forty-eight, all the trends, all the buzz around events like SBS, and distill it down into what really matters for you.

00:00:44: I think that's important because the feeling we're getting isn't about If this is all going to happen, it's about how fast.

00:00:49: Right.

00:00:49: The speed of implementation.

00:00:50: Exactly.

00:00:51: We're seeing huge shifts toward agentic AI, a heavy reliance on edge platforms, and this non-negotiable push for interoperable standards.

00:01:00: The era of the proof of concept is over.

00:01:02: We are firmly in the deployment phase.

00:01:04: We really are.

00:01:05: OK,

00:01:05: let's start at the top then.

00:01:06: AI and software-defined automation, the intelligence layer.

00:01:11: And what's fascinating is the biggest insight wasn't about some amazing new AI capability.

00:01:16: No, it was about the shocking lack of preparation.

00:01:19: Yes.

00:01:19: It's his infrastructure paradox.

00:01:21: Michael Finocchiaro put it beautifully.

00:01:23: He said, most manufacturers are basically set up to fail at AI because they skip the foundational steps.

00:01:29: They want the flashy result without the groundwork.

00:01:31: He called it the AI factory hierarchy of needs.

00:01:34: It starts with level one, data standards.

00:01:36: Level two, process digitization.

00:01:38: But here's the number that just floored me.

00:01:40: The

00:01:40: seventy percent.

00:01:41: You really, seventy percent of manufacturers don't even have those basic data standards in place.

00:01:46: Think about what that actually means for a second.

00:01:48: If your CAD files are a mess or your PLM system isn't talking to your ERP system, you have no single source of truth.

00:01:55: None.

00:01:55: So you can't possibly jump to level five, the autonomous agents level.

00:01:59: The AI has nothing reliable to work with.

00:02:02: The structure has to come before the sophistication.

00:02:04: But once that structure is in place, then the evolution of agentiKI is just genuinely exciting.

00:02:10: We're moving away from that rigid, you know, if-this-then-that programming.

00:02:15: We are.

00:02:15: We're moving towards these collaborative reasoning systems that can adapt on the fly.

00:02:19: So what does that look like in practice?

00:02:21: Well, Merck Jurzak mentioned a demo at the Siemens booth that really showed it off.

00:02:26: A simple chat prompt, just a natural language command.

00:02:29: Like

00:02:29: typing a sentence.

00:02:30: Exactly.

00:02:31: and it orchestrated autonomous co-bots to pick parts, assemble a little figurine, and then hand it over.

00:02:37: That's adaptive intelligence.

00:02:39: The system figured out the steps.

00:02:40: It didn't just follow a fixed code.

00:02:42: That sounds like a powerful coworker, but if I'm running a factory, that system is controlling millions in assets.

00:02:49: The safety stakes are huge.

00:02:51: How do you trust an AI that's reasoning on its own?

00:02:54: Trust is the absolute mandate.

00:02:55: It's everything.

00:02:57: And Victor M really emphasized this.

00:02:58: Industrial settings cannot operate with black boxes.

00:03:01: You need full explainability.

00:03:02: We need discipline, not just inspiration.

00:03:05: Precisely.

00:03:06: And that's leading to this push for what's being called Glassbox AI.

00:03:10: Glassbox

00:03:11: AI.

00:03:11: It's a hybrid.

00:03:12: It combines symbolic logic, machine learning, and LLMs.

00:03:16: But the key is the AI acts as an industrial co-pilot.

00:03:20: It augments human decisions.

00:03:22: It doesn't replace them.

00:03:23: So

00:03:23: every decision is auditable.

00:03:25: Fully explainable, fully auditable.

00:03:27: Which, for critical processes, is, well, it's non-negotiable.

00:03:30: Right.

00:03:31: And this new intelligence layer completely changes the software running the factory, like the MES.

00:03:36: Oh, absolutely.

00:03:37: Michael Sood highlighted that the whole MES market is moving faster toward cloud, deeper AI and IoT integration, and using the edge to process data locally.

00:03:46: And there's a new mandate being added.

00:03:48: Yeah,

00:03:48: this part's interesting.

00:03:49: embedding sustainability metrics right into the standard batch ports now.

00:03:53: So efficiency is now explicitly tied to environmental impact.

00:03:56: And to link that intelligence to the physical machines, you need a whole new architecture.

00:04:01: We saw a lot about software-defined automation using IEC-SX- one-four-nine-nine.

00:04:06: This is a huge paradigm shift.

00:04:08: IEC-SX- one-four-nine-nine lets you decouple the control logic, the software instructions, from the physical hardware.

00:04:14: So you can update the machine's brain without replacing the whole machine?

00:04:17: Basically, yeah.

00:04:18: It's like updating an app.

00:04:19: It allows for reusable function blocks, it avoids vendor lock-in, and it makes system upgrade so much faster.

00:04:26: Okay, so if the intelligence is getting that flexible, we need tools to test it before we deploy it.

00:04:32: And that brings us right to digital twins.

00:04:35: For a long time, this felt like something only for the giant automakers.

00:04:39: Right, it felt inaccessible.

00:04:41: How is that technology getting democratized for, say, a smaller midsize business?

00:04:47: Arthi Sariman had a great take on this.

00:04:48: She framed the digital quen as a powerful decision tool, specifically for smaller plants.

00:04:54: The key is you don't try to model your entire business at once.

00:04:57: You do it in strategic phases.

00:04:59: So where do you start?

00:05:00: What's the entry point for immediate value?

00:05:02: Stage one is the engineering simulation twin.

00:05:05: You use it for virtual commissioning.

00:05:07: So you're validating the layout, the logic, the safety.

00:05:10: All of it.

00:05:11: You do it virtually before you install any physical equipment.

00:05:14: It completely de-risks a new automation cell.

00:05:17: Then, stage two is the operational performance twin, which uses real-time data.

00:05:21: And that's where you find the quick wins.

00:05:23: Exactly.

00:05:23: You can run what-if scenarios, test schedules, find bottlenecks, all without stopping production.

00:05:29: And that ability is incredibly valuable, especially when you're retrofitting older facilities.

00:05:34: That retrofitting of Brownfield sites was a huge topic.

00:05:36: It was.

00:05:37: Rahul Gargan and Todd Mavralis both noted it's the most practical path to efficiency for so many companies.

00:05:43: And the digital twin is central.

00:05:45: You seemingly lead the new Kobach or the new AI process in a virtual world first.

00:05:49: You avoid the costly mistakes.

00:05:51: You

00:05:51: avoid the downtime.

00:05:53: And to make this happen, we saw new offerings aimed right at that smaller manufacturer.

00:05:58: Siemens introduced your SMB production optimization starter pack.

00:06:01: What's the angle there?

00:06:02: It directly targets the pain points workforce shortages.

00:06:06: High downtime.

00:06:07: Tara Young and Ronnie Hendrikak presented it as an affordable, scalable solution as a service.

00:06:13: So no massive upfront capital investment.

00:06:15: Right.

00:06:15: You don't need a huge specialized IT team.

00:06:17: It shifts the burden of expertise.

00:06:19: That's

00:06:20: crucial.

00:06:20: Okay, let's pivot from the software and intelligence to the physical stuff it controls.

00:06:25: Robotics and advanced motion.

00:06:28: There's a subtle debate going on about even the basic strategy

00:06:32: here.

00:06:33: Yeah, the custom versus general purpose robot debate.

00:06:35: Chris Sturgy weighed in and he cautioned against just defaulting to a general purpose robot because it's on the shelf.

00:06:41: Why though?

00:06:42: Why except the complexity of custom if you can buy off the shelf?

00:06:46: Because customization optimizes the application perfectly.

00:06:49: A general-purpose robot might have limitations in payload or speed or stroke that make it inefficient for a very specific task, like intricate bin picking.

00:06:59: If you need absolute precision and throughput, tailoring the hardware is often the better long-term investment.

00:07:05: While that debate is happening, the most visible trend is just the acceleration of humanoids.

00:07:09: The hype is turning into actual deployment plans.

00:07:11: Oh, big time.

00:07:12: We saw a huge movement here, especially out of Asia.

00:07:14: The Xiaomi CEO, Lei Jun, announced plans to mass introduce humanoids into their factories within five years.

00:07:22: Five years for full assembly lines.

00:07:24: And separately, Foxconn is partnering with Nvidia to develop AI-powered humanoids.

00:07:30: Peter Farke has highlighted that they're going to deploy them in US factories, specifically to build AI servers.

00:07:36: Wow.

00:07:37: But introducing humanoids next to existing machines, co-bots, human workers, that sounds like an integration nightmare.

00:07:44: It's the core challenge.

00:07:45: Jethyn Tellis stressed that digital simulation tools like Siemens Process Simulate are absolutely essential here.

00:07:51: You have to model the whole ecosystem.

00:07:53: The whole thing.

00:07:54: You check cycle times, safety envelopes, human ergonomics, all of it.

00:07:59: Before you commit to a physical layout, you have to.

00:08:02: It's just so stark to contrast that ambition with the policy discussions we're seeing in the West.

00:08:07: David Mills pointed out the UK is lagging.

00:08:09: Lacking behind countries like Slovenia and Slovakia even.

00:08:12: He's calling for stronger long-term policy support.

00:08:15: And that policy instability really stands out when you look at the global competition.

00:08:19: It does.

00:08:21: Dr.

00:08:21: Pascal MV gave this context about the speed and scale in China.

00:08:25: He sees it as them building AI armor against potential trade conflicts.

00:08:29: With companies like Medea Group using AI as a centralized nervous system for the whole operation.

00:08:34: And the numbers are just, well, they're staggering.

00:08:36: Fox kind of is pushing for lights out factories.

00:08:39: China installed nearly half the world's industrial robots last year alone.

00:08:42: That's an incredible rate of adoption.

00:08:44: And the hardware is keeping pace.

00:08:47: On the motion control side, the new Dynamics S-II-II drive system was a highlight.

00:08:52: Cedric Bartonhagen noted its modular design can reduce cabinet width by up to fifty percent.

00:08:58: Which is critical for space.

00:08:59: Critical space.

00:09:00: And it has integrated energy monitoring.

00:09:03: We also saw enomotics showcasing IE-six efficiency motors, which, as Milan Rajtar pointed out, is all about that immediate need to cut consumption.

00:09:11: So we have the software, the hardware, the intelligence,

00:09:14: but

00:09:14: none of it works without a common language.

00:09:16: Let's talk connectivity standards and bridging that huge gap between IT and OT.

00:09:21: This ITOT convergence is the foundation for everything.

00:09:25: predictive now.

00:09:26: Francisco Javier Franco Espinosa talked about combining platforms like Siemens Industrial Edge with the Azure Cloud.

00:09:33: So Edge on the factory floor, Cloud for the big picture.

00:09:36: Exactly.

00:09:37: Edge handles the low latency control locally and Azure provides the global intelligence and enterprise-wide visibility.

00:09:44: It bridges those two worlds.

00:09:46: And the key to making this multi-vendor global ecosystem actually function is standard.

00:09:52: It's always standards.

00:09:53: Andrea's faith emphasized that OPC UA companion specifications are like the critical grammar for this communication.

00:10:02: They ensure everyone is speaking the same language.

00:10:04: And building on that is the asset administration shell or AS.

00:10:09: Okay,

00:10:09: tell us more about the AS for you listening.

00:10:11: Why is this so important?

00:10:12: Well, Alessandro B, an axle with a bobbin, highlighted it as being like the digital passport for every single asset.

00:10:18: A digital ID.

00:10:19: A

00:10:19: digital ID that creates a common language.

00:10:21: It standardizes how an asset describes itself, its capabilities, its maintenance history.

00:10:25: This is what enables truly connected supply chains, not just connected machines.

00:10:29: That digital harmony is key.

00:10:31: OK, last big question.

00:10:32: Who owns this colossal transformation?

00:10:34: Yeah, who's in charge?

00:10:35: We know you need internal buy-in, but can you just outsource the whole change management process?

00:10:40: Jeff Winter's analysis says, basically, no.

00:10:43: Affirm, no.

00:10:45: His findings show a clear internal ownership model.

00:10:48: Fifty-two percent of manufacturers have created centralized teams specifically to drive these initiatives.

00:10:54: And the leadership is split right down the middle, fifty-one percent from operations, thirty-eight percent from tech.

00:11:00: The lesson is really clear.

00:11:01: You can outsource the implementation, you can hire consultants for strategy, but you cannot outsource the transformation itself.

00:11:08: It has to be driven from within.

00:11:09: It has to be.

00:11:10: Trying to manage Glassbox AI and humanoids without that internal alignment is just a recipe for chaos.

00:11:17: That brings us to the end of this deep dive into the very busy weeks, forty-seven and forty-eight.

00:11:22: If you enjoyed this, new episodes drop every two weeks.

00:11:24: Also check out our other editions on digital construction and digital power tools.

00:11:28: And just as we wrap up, let's connect those last two points.

00:11:32: policy and implementation.

00:11:33: Scott Clark noted that policy instability continues to be a drag on growth in Western markets.

00:11:39: Yet at the same time, global competitiveness demands that massive, scalable implementation that Dr.

00:11:44: Pascal MV described happening in China.

00:11:47: So for you, the listener, the ultimate challenge isn't just about picking which AI or which robot to adopt.

00:11:53: It's about how you future-proof your own operations and culture.

00:11:56: Aligning those teams we just talked about.

00:11:58: Exactly.

00:11:59: Aligning them against both internal chaos and external geopolitical unpredictability.

00:12:04: And that is a problem no single technology can solve.

00:12:07: It takes leadership.

00:12:08: Leadership that understands standards and scale.

00:12:11: Precisely.

00:12:12: Thanks for joining us.

00:12:13: Subscribe to The Deep Dive for more analysis you can trust.