Best of LinkedIn: Smart Manufacturing CW 11/ 12

00:00:00: This episode is provided by Thomas Allgaier and Frennus, based on the most relevant LinkedIn posts about smart manufacturing in calendar weeks eleven and twelve.

00:00:09: Frenness is a BDB market research company that supports Smart Manufacturing providers with building feature-by-feature competitive intelligence That shows exactly how their product stacks up against the competition.

00:00:21: You can find more info In the description.

00:00:23: Let's get right into The Insights.

00:00:24: Okay let's unpack this.

00:00:26: We are cutting through the noise today to bring you the top smart manufacturing trends that have been dominating LinkedIn over the past two weeks.

00:00:33: Yeah, and there's a lot of noise out there

00:00:35: There really is.

00:00:36: I mean if you work in construction or manufacturing or heavy industry You know these hype cycles can just be exhausting.

00:00:43: Oh,

00:00:43: absolutely exhausting!

00:00:44: So today we are strictly looking at what is actually surviving the brutal reality of The Factory Floor?

00:00:50: Right and when you look at the curated data from the past two weeks a massive overarching theme just jumps out immediately.

00:00:57: Which Is What Exactly?

00:00:59: Well,

00:00:59: the market is shifting aggressively.

00:01:01: I mean we are moving away from these ambitious isolated experiments and pivoting hard into actual execution.

00:01:08: so The whole pilot phase has done basically

00:01:10: yeah Yeah that tolerance for what many call pilot purgatory.

00:01:14: You know where a company tests the technology forever but never actually deploys it at scale.

00:01:19: That's officially over manufacturers our demanding tools that solve immediate bleeding neck problems right now.

00:01:26: Which brings us perfectly to our first major topic, which is industrial AI.

00:01:30: Right

00:01:30: the big buzzword.

00:01:31: always we hear about artificial intelligence constantly but running a language model in a quiet air-conditioned office Is a fundamentally different sport than running it next to a fifty ton hydraulic press?

00:01:43: It's not even the same league.

00:01:44: Exactly and looking at the insights shared by Chris Stevens and Anne Chevrier recently A really clear consensus is forming here.

00:01:52: Yeah they both hit on something crucial.

00:01:54: The winning companies in manufacturing aren't the ones just throwing the biggest budgets at AI pilots.

00:02:00: They're the ones with strategic clarity on which specific bottlenecks they need to solve first.

00:02:06: Context is completely beating flashy dashboards.

00:02:10: Chevrolet made a really compelling point about this, actually she noted that most companies you know...they respond to the pressure launching these disconnected initiatives.

00:02:21: Like

00:02:21: just trying to check a box?

00:02:22: Exactly, so you might see a predictive maintenance proof of concept on one line and then a connected machine dashboard on another.

00:02:29: And she calls this activity without architecture

00:02:32: Activity Without Architecture.

00:02:34: I like that.

00:02:34: it's

00:02:34: so accurate.

00:02:36: I mean, a dashboard might look great to a visiting executive but it just displays beta.

00:02:40: It doesn't actually make the decision or change of physical outcome on the floor.

00:02:43: Right and i really have push back this whole industry narrative of plugging AI into legacy systems

00:02:51: The plug-and play myth.

00:02:52: Yes!

00:02:52: I hear that pitch all time And feels like trying drop high performance Formula One engine Into heavy duty agricultural tractor without upgrading transmission It's

00:03:03: gonna tear itself apart

00:03:04: Completely.

00:03:04: you cannot put high-speed, predictive decision making software into a facility that still runs its actual physical workflows on paper or operates with the two week delay in supply chain data.

00:03:17: The physical reality just can't keep up Exactly.

00:03:21: And the underlying reason for that disconnect is what Dina Kaur Ramamurthy highlighted in his recent post, he pointed out that AI and manufacturing rarely fails because the actual underlying models are weak

00:03:32: right?

00:03:32: The tech itself is usually fine.

00:03:33: Yeah!

00:03:34: The math works.

00:03:34: it fails Because of What He Terms Sociotechnical Constraints.

00:03:38: Okay let's break That Down For The Listener.

00:03:40: Socio-technical constraints yeah.

00:03:42: so we're talking about the intersection Of The Machine's Physics...and The Human's Behavior.

00:03:46: Yes

00:03:46: exactly..so on the technical side you have real-world physics and tack time.

00:03:51: Tack

00:03:52: Time being the rhythm of a line.

00:03:54: Right, TACK TIME is the precise rhythm at which product needs to be completed to meet customer demand

00:04:00: Makes sense.

00:04:01: So if an AI model suggests slowing down machine by maybe five percent preserve tool life but doing that throws off the tack time for entire assembly line downstream,

00:04:12: then it's a completely useless recommendation.

00:04:14: Exactly!

00:04:15: It doesn't understand physical context.

00:04:17: and on social side you have operator trust

00:04:20: which is huge

00:04:21: massive.

00:04:21: If an AI system does not understand operational rhythm of plant or if makes one glaringly obvious mistake The human operators are just going to hit the override button and they will ignore it permanently.

00:04:34: Yeah,

00:04:34: They'll just go back to their manual way.

00:04:36: And if operator trust is the bottleneck then Muserat Hussein's recent post about cloud AI versus edge AI Is absolutely critical here.

00:04:44: Oh that was a great pose

00:04:45: It really was.

00:04:46: He brought up this concept of vibe coding where people get incredibly excited About autonomous AI agents Just building software apps in ten minutes using the cloud

00:04:55: Which is great for software.

00:04:56: Sure its amazing for software.

00:04:58: But if you are managing a plant floor today and relying on cloud AI, Hussein asks a really vital question.

00:05:07: What happens when you bring that cloud dependency into a steel mill?

00:05:10: Right the stakes transition from digital inconvenience to actual physical danger.

00:05:15: Exactly

00:05:16: A hallucination or drop connection in pieces of software is just funny bug You know.

00:05:21: refresh page move-on

00:05:22: No harm done.

00:05:23: But if your machine is spinning at ten thousand RPMs, a five hundred millisecond delay in detecting an anomaly because the system had to you know ping a server and another state that isn't a glitch.

00:05:33: No it's a catastrophic breakdown.

00:05:35: A supply chain halts?

00:05:37: A multi-million dollar asset gets destroyed or worse someone actually get seriously injured.

00:05:41: And thats why Hussein conclusion Is that heavy industry desperately needs edge gen AI.

00:05:47: The AI cannot live In distant data center.

00:05:49: They have to be local

00:05:50: Right.

00:05:50: The inference, the actual decision-making process has to happen at the edge physically right there on machine's local hardware.

00:05:57: So you completely eliminate latency?

00:05:59: Exactly.

00:06:00: And by doing that, you empower what Hussein calls Operator Five Point-O.

00:06:04: Operator

00:06:05: five point oh?

00:06:06: I really love that framing!

00:06:07: It's

00:06:07: good.

00:06:07: right because You are giving the human on the floor zero latency highly contextual decisions.

00:06:13: Because ultimately no matter how good the AI gets The Human is the final failsafe On a factory floor.

00:06:19: Right it shifts the narrative.

00:06:20: We constantly hear That AI Is coming to replace the manufacturing workforce

00:06:24: All the time

00:06:25: But the data shows something entirely different.

00:06:27: AI isn't replacing these workers, it is frantically trying to back them up before we hit a massive demographic cliff.

00:06:34: Yes and this ties directly into the concept of industry.

00:06:37: five point oh right if industry.

00:06:39: four point Oh was all about machine connectivity in raw data.

00:06:42: Industry.

00:06:43: five point O places are heavy.

00:06:45: human centric focus backs onto the manufacturing floor.

00:06:48: Bringing The Human back into the loop?

00:06:49: Exactly!

00:06:50: There was a fascinating debate recently between John Healy and Jeff Winter on LinkedIn about this, they were discussing whether Industry Five Point-O is truly new distinct era or just an necessary evolution of Industry.

00:07:01: Four Point-Oh.

00:07:02: And

00:07:02: what's the verdict?

00:07:03: Well regardless of which label you slap it...the core operating principle remains the same Technology must augment worker judgment not attempt to blindly override it.

00:07:14: We desperately need that augmentation right now.

00:07:16: Sean C shared some statistics that should frankly keep any plant manager awake at night.

00:07:21: Yeah, those numbers were rough

00:07:23: very rough.

00:07:24: by twenty thirty three two point eight million manufacturing workers are retiring.

00:07:28: Wow and here's the terrifying part They're taking seventy percent of the undocumented tribal knowledge with them when they walk out the door.

00:07:38: Seventy percent?

00:07:39: That's just, that's staggering!

00:07:41: It is...the little tricks and workarounds.

00:07:43: you know knowing this sound a machine makes right before it jams..that its all just vanishing.

00:07:47: And

00:07:47: When You Lose seventy percent of your institutional memory The financial impact is immediate.

00:07:51: Seeing noted This specific brain drain costs organizations roughly forty seven million dollars a year.

00:07:57: Forty Seven Million Yeah

00:07:58: ...and comes in form of increased scrap rates extended training periods for new hires and duplicated problem solving.

00:08:04: What do you mean by duplicated problem solving?

00:08:06: Well, it's where a new engineer spends say three days fixing the machine issue that their retired veteran used to fix in ten minutes just with a tap of a wrench.

00:08:16: Right because they knew the quirk of this specific machine

00:08:19: Exactly!

00:08:20: And the demographic replacing them isn't sticking around to learn at the hard way either.

00:08:24: Say he pointed out that Gen Z's average tenure and manufacturing is one point-one years Barely over a year.

00:08:31: Yeah They were rotating out before they even have a chance to build up that kind of intuition.

00:08:35: Which is a huge

00:08:36: problem!

00:08:37: So how's the industry actually solving this?

00:08:40: I mean, How do you digitize the intuition Of a thirty year veteran Before their retire?

00:08:45: Manufacturers are turning To RAGGI based AI systems.

00:08:49: Okay, right.

00:08:50: Yeah our AG stands for Retrieval Augmented Generation.

00:08:54: Instead of just asking A generic public language model question RJeg connects the AI directly to a company's secure internal knowledge basis.

00:09:02: I was looking at their own stuff!

00:09:03: Right, we are talking about decades of scattered PDF manuals maintenance logs historical process data and even transcribed interviews with those retiring veterans.

00:09:13: So when a twenty-two year old operator encounters some obscure fault code on a legacy machine They don't have to dig through a dusty filing cabinet somewhere.

00:09:23: Exactly, they ask the RRAG system which instantly retrieves this specific internal document from maybe ten years ago where the retiring veteran documented the exact

00:09:34: fix

00:09:35: and it generates clear step-by-step instruction based purely on that retrieved verified company data.

00:09:42: It turns static triangle knowledge into dynamic interactive query.

00:09:46: Here's what gets really interesting.

00:09:48: Peter Wexer shared a real-world success story about this dynamic playing out at Schneider Electric's Wuhan factory.

00:09:54: Right,

00:09:55: I saw that!

00:09:55: They were facing massive talent crisis.

00:09:58: Their technician turnover rate had skyrocketed to forty eight percent.

00:10:01: Forty

00:10:01: Eight percent is basically revolving door.

00:10:03: Yeah.

00:10:04: furthermore only about twenty percent of their employees were adequately skilled to handle the new layers of automation they are bringing into facility.

00:10:11: And a forty eight per cent turn over rate in high tech facilities means you're constantly on boarding.

00:10:16: You can't get ahead.

00:10:17: No, you can never achieve operational excellence because half your workforce is always just learning the basics.

00:10:23: But they turned it around!

00:10:24: They deployed GenAI for assisted maintenance and competency management.

00:10:29: And specifically...they use what we call agentic AI.

00:10:32: Agentic AI?

00:10:33: Yeah

00:10:34: For a listener when say agentic A-I.

00:10:36: We mean The software isn't just passive search bar waiting for an operator to type of question.

00:10:42: It acts as independent agent

00:10:44: Meaning its proactive

00:10:45: Exactly.

00:10:46: It actively monitors what a worker is doing, maps their skill gaps in real time and pushes personalized training or guidance to them before they even realize that they are stuck.

00:10:57: And the results of those implementations speak for why turnover happens at first place.

00:11:02: by deploying this... Schneider Electric dropped that forty-eight percent turnover rate down to just

00:11:07: six percent.

00:11:08: It's a massive drop,

00:11:09: and they boosted their workforce readiness from twenty percent to seventy-six percent.

00:11:13: Think

00:11:13: about the psychology there.

00:11:14: I mean.

00:11:15: why did turn over drops so drastically?

00:11:17: because they weren't frustrated anymore

00:11:19: right?

00:11:19: Because in new hire is no longer spending half Their shift terrified of breaking a multi million dollar machine while flipping through a dense paper manual.

00:11:28: The

00:11:28: frustration has completely gone.

00:11:30: The AI assistant gives them the answers they need instantly, their confidence goes up.

00:11:35: They become confident faster and suddenly...they don't want to quit!

00:11:39: It's technology deployed as a retention

00:11:41: tool.".

00:11:43: So we've solved the brain drain by capturing human knowledge.

00:11:47: but knowing how to diagnose a machine doesn't matter if you have enough human hands physically on the floor to move material or turn the wrench.

00:11:58: which is why the robotics race, fundamentally changing equation right now.

00:12:02: The robots are finally stepping out of viral video stage where they do backflips in a controlled lab and onto the messy human design factory floor.

00:12:10: It's about time

00:12:12: Dale Tutt made fascinating observation regarding this sudden explosion of humanoid robot.

00:12:16: specifically

00:12:17: What did he say?

00:12:18: He pointed it that the massive commercial appeal of a bipedal humanoid robot isn't just that it looks cool or futuristic, It's purely about infrastructure.

00:12:28: Form follows function!

00:12:29: Exactly

00:12:30: I mean for the last century we have designed every factory Every staircase every tool and every workstation around The dimensions and limitations Of the human body.

00:12:39: yes

00:12:40: you buy A traditional robotic arm on a track.

00:12:43: You had to tear out your conveyor belts?

00:12:47: put up massive safety cages.

00:12:49: It's a massive facility redesign, it costs a fortune!

00:12:52: Right but if you buy a humanoid robot... ...it can walk down the same narrow aisle of human walks-down.. ..It can pick up that standard drill a human uses and drops directly into the existing production line.

00:13:04: The barrier to entry just dropped almost zero.

00:13:07: While the form factor is exciting, Nikhil Chathary offered a crucial pivot on this... ...focusing on the geopolitical and scaling realities of this shift.

00:13:14: Oh!

00:13:14: This is important.

00:13:15: Very He stated flatly that the robotics race won't be one by whoever has the smartest dancing robot demo on YouTube.

00:13:22: It will be One By Supply chains.

00:13:24: We are moving from prototype-to-production And now into infrastructure

00:13:28: Building its scale.

00:13:29: Exactly, if you want to deploy a million humanoid robots who actually has the component ecosystems and manufacturing capacity To build a million high torque actuators in specialized batteries.

00:13:40: And this is where Europe Actually Has A Massive Often Overlooked Advantage.

00:13:45: Yeah Oli Kansyovicci Made A Really Bold Provocative Claim Last Week About This.

00:13:50: What

00:13:50: Was The Claim?

00:13:50: He

00:13:51: Posted That Saus Is Dead that

00:13:53: It's Quite The Statement In An Era Dominated By Software.

00:13:55: Well I Know Right But he argues that if you build software today, a massive tech giant might just update their model tomorrow and completely swallow your entire feature set.

00:14:05: True The digital space is infinitely replicable but physical automation building complex robots at scale That requires decades of automotive engineering deep established supply chains And precision manufacturing know-how

00:14:18: Things You Can't Just Spin Up On A Server.

00:14:20: Exactly.

00:14:21: Yeah, so Vici argues that Europe's massive opportunity lies in pivoting its legacy automotive and industrial heritage directly into robotics because you just can't code a supply chain overnight.

00:14:32: Europe

00:14:32: certainly has the industrial foundation for it, however Alexei Bermistrov and Karsten Heizer both provided an necessary reality check regarding actual execution?

00:14:42: Oh

00:14:43: Yeah they pointed out that China is currently setting an incredibly aggressive pace for industrialized scaling specifically in robotics and additive manufacturing.

00:14:51: Right.

00:14:51: Burmistrov noted that while the West might be exceptional at conceptualizing these frontier ideas and building the first incredible prototype, China is quietly surpassing US and Europe in actual deployment speed and cost-efficient scaling.

00:15:05: Heuser used a brilliant analogy here.

00:15:07: actually... What was it?

00:15:08: ...he compared China to a marathon pacemaker In industrialized additive manufacturing.

00:15:14: China is the pacemaker setting the speed for the rest of the world.

00:15:17: They have full supply chain coverage from raw materials to final assembly, and they are moving relentlessly

00:15:23: fast.".

00:15:23: So if Western manufacturers want a stay in their race... ...they can't just conceptualize on the lab?

00:15:28: No!

00:15:29: They have to execute and deploy on the floor….

00:15:31: But here's the tension with that – If global competition moves incredibly quickly, manufacturers cannot afford trial-and-error on physical shop floors.

00:15:41: Oh definitely not.

00:15:42: You cannot just halt a running production line for three days to see if the new robot configuration improves your throughput.

00:15:49: If you are wrong, you bleed money!

00:15:51: You

00:15:51: have to perfect physical changes in virtual world first...

00:15:54: Exactly

00:15:55: What's fascinating here is how digital twins matured to solve exactly that problem For long time.

00:16:01: when companies said they had a digital twin it basically meant they had fancy CAD visualization of their factory.

00:16:08: Just nice picture.

00:16:10: It looked nice, but it didn't do much.

00:16:12: Today a true digital twin is core decision-making infrastructure!

00:16:25: Pepsico realized they needed to modernize several of their older facilities, but instead doing physical pilot programs and moving heavy equipment around the floor to see what worked.

00:16:40: They used Siemens Digital Quinn Composer integrated with NVIDIA on Diverse.

00:16:45: So they build a high fidelity virtual replica of their facilities.

00:16:49: Yes, we are talking about simulating every single machine the speed of every conveyor belt The exact flow of pallets and even the walking paths of human operators.

00:16:59: And then let AI run massive simulations of facility upgrades inside that virtual environment.

00:17:05: Just testing everything virtually.

00:17:07: Imagine being able to simulate a palette flow crashing into bottle neck ten thousand times in a virtual world Just to find the optimal conveyor angle.

00:17:16: That's amazing!

00:17:17: By

00:17:17: doing this, PepsiCo identified upto ninety percent of potential operational issues before a single physical change was made on the actual floor.

00:17:25: The risk reduction there is just incredible.

00:17:27: and when you remove the guesswork like that?

00:17:29: The financial impact is undeniable.

00:17:31: Absolutely.

00:17:32: by solving those bottlenecks virtually PepsiCo achieved a twenty percent increase in actual physical throughput once the changes were implemented.

00:17:41: Wow!

00:17:41: Furthermore, they reduced their capital expenditure by ten to fifteen percent

00:17:45: because they weren't buying the wrong stuff

00:17:46: exactly.

00:17:47: yeah Because these simulated the physics beforehand.

00:17:49: They knew exactly what equipment to buy and exactly where to put it before they spent a single dollar.

00:17:55: that

00:17:55: is huge.

00:17:57: And you know, if we connect this digital twin concept back to the workforce crisis we discussed earlier these simulations are also revolutionizing how we train in next generation.

00:18:07: How so?

00:18:07: Well, Joss Lichter posted about a fascinating partnership between Heller and Siemens.

00:18:12: They are using digital twins alongside of physical scaled-down functional five axis CNC machine called the Token training machine.

00:18:20: Wait so it's real machine that actually cuts metal but its used purely for training?

00:18:24: It is a real functioning machine yes But it has seamlessly paired with its own digital twin.

00:18:29: Oh I see.

00:18:30: And this completely de-risks the learning curve for that incoming Gen Z workforce.

00:18:36: An apprentice can learn complex programming, a machine setup and precision machining workflow safely in virtual environment first.

00:18:43: So if they mess up,

00:18:44: If make a math error and crash the virtual spindle into the virtual metal at high speed nobody gets hurt.

00:18:51: And no expensive tools are destroyed.

00:18:52: That is so smart.

00:18:53: They learn from mistake perfect their code only then do move to physical talking machine.

00:19:00: So what does this all mean?

00:19:02: When you layer these themes together, AI moving to the edge, ROG systems capturing human knowledge.

00:19:08: Scalable robotics taking on physical tasks and digital twins simulating the future.

00:19:14: What is the ultimate end game for a manufacturing executive here?

00:19:17: according to Alex Allison The ultimate goal of stacking All This technology boils down To one singular metric which Is speed specifically the speed Of iteration.

00:19:27: how fast can You adapt right?

00:19:29: Allison points out that the winning firms are the ones utilizing what we call digital threads.

00:19:33: Digital Threads Yeah,

00:19:34: and for the listener a digital thread is an unbroken flow of data That connects the initial digital design of a product through its physical manufacturing all The way to it's real-world performance in the hands of the customer.

00:19:47: It connects the whole life cycle

00:19:48: exactly Allison notes.

00:19:50: at the best companies use these digital threads to trace a field failure identify the root cause in the factory, implement an engineering fix and push that fix under the production floor all before their competitors even finish processing the customer's return authorization.

00:20:04: They are closing the loop almost instantly.

00:20:07: Instantly!

00:20:08: And that brings us to the harsh reality of where the manufacturing industry sits today.

00:20:12: Jacobo Lorette-Casal leaves with a very sobering thought regarding this whole transition.

00:20:17: Yeah

00:20:17: his post was intense.

00:20:19: It was He points out for last decade Industry four point oh was sold to manufacturers primarily as an innovation and growth story, right?

00:20:28: It was pitched as a luxury.

00:20:30: You know A way to get a little more efficiency or a bit more visibility for the shareholders.

00:20:35: Nice to have upgrade when budgets are loose

00:20:37: Exactly.

00:20:38: but the macroeconomic environment has fundamentally shifted with The massive energy shocks Europe has faced combined with ongoing relentless supply chain volatility.

00:20:50: Industry four point zero is no longer a growth story, what does it do?

00:20:54: It has to survival story.

00:20:55: Cassel notes that the factories bleeding margin today aren't the ones that took a risk and bought the wrong technology.

00:21:01: they're The ones

00:21:03: who just waited for the perfect flawless solution, while the world completely changed around them.

00:21:09: Exactly!

00:21:09: If you are running a plant today and have no real-time energy consumption data No production visibility And don't know where your physical bottlenecks are until a spreadsheet arrives three weeks later You're operating blind in highly volatile environments.

00:21:24: He can't steer the ship.

00:21:26: The lack of real-time data is no longer just an IT problem for the engineers to figure out.

00:21:31: It's a board level existential risk.

00:21:38: If you haven't upgraded your core data infrastructure and workforce enablement, the engine of global competition is going to tear you apart.

00:21:51: You just can't survive on a legacy momentum over the last twenty years!

00:21:55: It really isn't existential risk which leaves us with one final thought to mull over today before we go.

00:22:01: If

00:22:03: AI and digital twins are perfectly capturing the expertise of your veteran workers, an advanced robotics are optimizing every physical movement on your floor.

00:22:12: What happens to the competitive advantage of your company when every other factory in the world buys the exact same technology?

00:22:19: Oh man... if the software and robots are equally available everyone The only true differentiator left might just be creativity and adaptability for the physical humans running it all.

00:22:29: That is a fascinating point.

00:22:32: If you enjoyed this episode, new episodes drop every two weeks.

00:22:35: Also check out our other editions on digital construction and digital power tools.

00:22:38: Thank You so much for joining us in the deep dive.

00:22:41: Don't forget to subscribe So that you can stay ahead of the curve as industry continues to evolve Until next time.