Best of LinkedIn: HANNOVER MESSE 2026

00:00:00: This episode is provided by Thomas Allgeier and Frennis, based on the most relevant LinkedIn posts about handover mess.

00:00:07: Frennes is a B to B 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.

00:00:18: You can find more info in the description.

00:00:20: Alright.

00:00:20: so I think we're ready.

00:00:21: jump in.

00:00:22: Yeah absolutely.

00:00:23: So imagine a humanoid robot right?

00:00:26: And it makes this real-time decision to halt a production line, but not because it hit some pre-programmed hard stock.

00:00:34: But because it actively perceived this really subtle vibration in the motor that a human totally missed analyzed their risk and just adapted its behavior on fly

00:00:43: Which sounds like science fiction.

00:00:44: honestly

00:00:45: Exactly!

00:00:46: This isn't some twenty thirty road map vision...this was live running code at Hanover Mess this week If you're listening here navigating realities of construction and manufacturing industry.

00:00:57: So today we're bringing you a highly curated dip dive.

00:01:00: Yeah, We've taken this massive stack of insights specifically the raw dispatches and LinkedIn debates from people actually walking in halls of Hannover.

00:01:10: Messing

00:01:10: Right.

00:01:10: our mission is to extract critical trends shaping factory floor no fluff just grit.

00:01:17: smart build and manufacturing And

00:01:19: honestly it's perfect time for that.

00:01:21: I mean, HandoverMessage is always a bellwether.

00:01:24: But reading through the analyses from engineers and executives.

00:01:27: this year there's this distinct shift in global tone.

00:01:31: Yeah you can definitely feel that

00:01:33: Like the era of abstract tech promises and isolated lab experiments?

00:01:38: That's over.

00:01:38: We are moving heavily into concrete operational execution.

00:01:42: Okay let's unpack because it was set up for our first major theme which is this execution reality or well What I'm calling the end of the AI pilot phase right?

00:01:53: It feels like we spent the last three years designing this flashy futuristic concept car.

00:01:58: We've admired it in the showroom.

00:02:00: We debated that theoretical specs Yeah, but this year at Hanover The industry was basically forced to answer How do we actually mass produce this car and drive it in real rush hour traffic?

00:02:11: exactly?

00:02:11: But concept car has to survive the pavement.

00:02:14: okay.

00:02:15: And you know Peter fendle From Capgemini captured this beautifully in his dispatch from the fair.

00:02:20: He framed it perfectly, observing that AI just swapped a lab coat for The Blue Collar.

00:02:25: I love that phrasing

00:02:26: Right But he brought the receipts to show how painful that transition actually is...he shared some compelling research data.

00:02:33: So right now Roughly seven out of ten industrial companies have at least one AI use case in production.

00:02:40: Okay, but fewer than three out of Ten I've actually managed to scale it across their operations.

00:02:45: Wait

00:02:45: less then three out ten.

00:02:46: Let me pause you right there because if everyone from the C-suite down is just throwing money at AI Why is there such a massive graveyard of pilot projects?

00:02:54: What's the actual bottleneck preventing that scale?

00:02:56: well?

00:02:57: Well, I mean It comes down to the friction of the physical world.

00:03:00: yeah pilot program It lives in this pristine structured sandbox.

00:03:04: The data is clean, the variables are controlled but scaling it across a noisy complex legacy heavy manufacturing floor is a totally different engineering challenge.

00:03:16: you're dealing with machines from the nineteen nineties that don't speak modern protocols.

00:03:21: oh yeah You've got fluctuating environmental conditions undocumented legacy code and Fintl made a really compelling argument that bridging gap moving from the three to this seven is exactly where profit and loss impact.

00:03:45: And you can see the major players aggressively restructuring to bridge that gap.

00:03:49: like Nikhil Rajagopalan at Accenture sparked a big discussion when he pointed out that AI is no longer viewed as this future ambition, right?

00:03:56: It has fundamentally become the core operating fabric of engineering in manufacturing.

00:04:06: Their CEO, Christian Klein was on stage with the German Chancellor committing twenty billion euros to AI and cloud infrastructure in Europe.

00:04:14: Twenty billion is a staggering figure even for SAP!

00:04:18: It's massive but what caught my eye wasn't just the dollar amount.

00:04:22: it Klein specifically focused on integrating this AI infrastructure into the middle stand, you know?

00:04:29: The medium-sized enterprises that actually form the backbone of the industrial economy.

00:04:34: Right so they aren't just building bespoke tools for massive multinational conglomerates anymore

00:04:38: Exactly!

00:04:39: They're embedding Industrial AI directly in to standard supply chain and service processes

00:04:44: Which logically forces us look at how this intelligence manifests physically.

00:04:49: Software living in a cloud server optimizing the supply chain is one thing, but manufacturing is inherently physical.

00:04:55: Yeah you're bending metal mixing chemicals

00:04:57: moving heavy pallets.

00:04:59: if AI Is The New Operating Fabric?

00:05:02: How does it interact with the physical world?

00:05:05: and the line between digital intelligence and physical execution essentially vanished at this year's fair.

00:05:11: You could literally see that on the floor.

00:05:13: Manish Goble from AWS had this great anecdote where he joked about actively being chased down the show floor by robots.

00:05:20: Physical AI was arguably the dominant visual of The Week, but we aren't talking about traditional robotic arms swinging blindly inside safety cages repeating the exact same spatial coordinates.

00:05:32: We are looking at robots that see learn remember and act dynamically in shared human spaces.

00:05:39: This shift from programmed repetition to dynamic perception is the crucial mechanism here.

00:05:45: But how does that actually work?

00:05:47: Well, Alexandre Embry from Capgemini showcased exactly how it works.

00:05:51: He highlighted a live humanoid system on their booth and this wasn't some scripted hard-coded demo... Okay You could change the task or alter rules of environment right in spot.

00:06:00: The underlying AI models allow robot to perceive new state understand your human intent and adapt its physical actions real time.

00:06:08: Wow Yeah!

00:06:09: The perception, cognition & motion are no longer these separate silos.

00:06:13: they operate continuously with single closed loop.

00:06:16: I mean that is the holy grail of robotics.

00:06:20: Yeah,

00:06:20: but we also have to ground this in the commercial reality.

00:06:23: Of say a procurement manager and Andreas reichert did a brilliant boots on-the-ground Reality check on this.

00:06:28: Oh what he do?

00:06:29: He bypassed all The flashy demos And went hunting for humanoid robots That a factory could actually buy right now For under one hundred thousand euros.

00:06:37: oh that Is a fantastic pressure test.

00:06:39: What did you find?

00:06:40: he found A stark market reality.

00:06:42: The Western alternatives were mostly unavailable, essentially telling buyers to check back later in the year.

00:06:49: Meanwhile Chinese vendors like Swinsoft and LIMAX are completely dominating this price point in availability.

00:06:56: they're ready to ship.

00:06:57: however Rykert pointed out a massive underrated physical constraint for these robots.

00:07:04: Autonomous battery swapping.

00:07:05: Wait

00:07:05: really?

00:07:06: Out of all the AI challenges batteries

00:07:09: Think about the mechanics of it.

00:07:11: An impressive humanoid with dynamic perception is completely useless if it dies three-quarters through a shift, and then requires a human technician to manually decouple or swap a heavy power unit.

00:07:24: in a dirty high vibration factory environment having a robot autonomously navigate perfectly aligned physical contacts with millimeter precision and swap a heavy battery pack without human intervention, it's incredibly difficult.

00:07:40: For real continuous shift operations power logistics remain massive hurdle.

00:07:46: I hadn't even considered the physical tolerances required for that.

00:07:49: But here's where it gets really interesting, we've been talking about physical robots acting as agents in the real world.

00:07:54: but there is another kind of agent taking over the engineering side and thats Agentec AI.

00:07:58: Let

00:07:59: us clearly separate this from the listener.

00:08:00: how does agentec ai functionally differ?

00:08:03: form the generative ai most people use every day?

00:08:06: So normal AI like a chatbot basically a reactive assistant You prompt it, it generates text or code.

00:08:12: And then you the human go copy that code verify and implement it.

00:08:16: right agenda AI doesn't just assist It acts autonomously to complete multi-step complex tasks.

00:08:22: a prime example That really blew up on LinkedIn came from Kevin O'Donovan and Dr Michael Schrap who analyzed Siemens launch of The Eigen engineering agent.

00:08:30: okay?

00:08:30: I saw some buzz about this.

00:08:31: yeah This agent doesn't Just give advice.

00:08:33: it autonomously writes the PLC Code and configures the TIA portal projects.

00:08:38: Siemens is claiming this delivers ready-to-use results that boost engineering efficiency by up to fifty percent.

00:08:44: Okay, I have to play devil's advocate here.

00:08:46: whenever i hear of Fifty Percent Efficiency Boost Claimed By A Software Vendor My Alarm Bells Ring!

00:08:52: Fair enough

00:08:53: How Is An AI Writing KLC Code Safely?

00:08:57: If an LLM hallucinates Python code for a website maybe a button doesn't work.

00:09:01: but if an Ai Hallucinates PLC Logic a multi ton press crashes into a die.

00:09:07: How do they prevent that?

00:09:08: That is exactly the right question to ask.

00:09:11: and it's why this as an engineering agent, not just a raw language model.

00:09:15: It operates within the highly constrained environment of the TIA portal.

00:09:19: Oh okay...It's NOT guessing at machine physics from scratch!

00:09:22: Its utilizing standardized Siemens libraries & existing digital twins to generate the structural logic.

00:09:28: The human engineer is elevated from writing boilerplate code for Scratch to basically verifying and fine-tuning the agents output that where the fifty percent efficiency comes

00:09:38: That makes much more sense.

00:09:39: Yeah, it's shifting the human from creator to editor

00:09:41: precisely and Catrin aren't from TK elevator brought another tangible example.

00:09:47: They physically brought a glowing elevator to the fair, to demonstrate how agentic AI is actively supporting their twenty-five thousand service technicians.

00:09:56: Oh wow!

00:09:57: Yeah

00:09:58: instead of a technician digging through hundreds of pages of PDF manuals while sitting on top of an elevator car The AI agents are diagnosing operational signals pulling the exact contextual maintenance history and making actionable decisions to support the repair.

00:10:14: That's incredible.

00:10:15: But here is the why all of these systems, whether it's a humanoid robot or an engineering agent are going to hit.

00:10:21: Yeah You can have the most advanced dynamic perception and the smartest eugenic AI in the world.

00:10:25: yeah but if those systems cannot access The right contextual data they're highly expensive paperweights which brings us To the hidden foundation Of everything we've discussed industrial Data architecture and interoperability.

00:10:38: Oh, absolutely.

00:10:38: It's like hiring a team of Michelin star chefs that your agentic AI.

00:10:43: but you lock them in kitchen where the pantry is completely disorganized half ingredients are hidden and locked cabinets all labels written in proprietary languages.

00:10:52: That great analogy.

00:10:53: it doesn't matter how good chef they can cook masterpiece if literally cannot find or recognize flower.

00:11:00: And that is a very accurate representation of modern factories IT stack.

00:11:05: What's fascinating here, how Yitzhak Kesselman from Microsoft articulated this perfectly during the fair.

00:11:12: he stated quite bluntly manufacturing doesn't have data problem.

00:11:15: it has connection problems operational signals real-time temperature of boiler vibration spindle those live isolated on factory floor.

00:11:25: But the business context, supply chain delays maintenance budgets raw material costs that lives in enterprise IT and finance.

00:11:32: Yeah these systems need to inform each other In real time for an AI make a good decision.

00:11:37: but those two worlds rarely speak The same language.

00:11:40: you essentially have a nervous system And a brain in the same body That just refused To communicate

00:11:45: exactly.

00:11:46: dr.

00:11:47: Adrienne Reich weighed-in on this noting that proprietary silo data ecosystems are the ultimate bottleneck.

00:11:54: He pointed out that the concept of The Digital Twin has evolved.

00:11:58: It's no longer just a static, three-D CAD model you look at on screen.

00:12:01: it is actively becoming the operational brain orchestrating the physical world faster than real time.

00:12:07: but our brain only works if the synapses fire and data flows freely across the entire value chain.

00:12:14: And the most encouraging part of Hanover Mess was seeing actual solutions to this, not just theoretical white papers.

00:12:20: Like Hannah Boonegarden highlighted an incredible live demonstration by a project called ChemX they were executing cross-industry data exchange using digital material passports.

00:12:31: For those unfamiliar how does a Digital Material Passport function practically?

00:12:34: Think about it as cryptographic proofs physical materials properties and origins that travels with the material digitally.

00:12:41: What boomgarden stressed was that this wasn't some PowerPoint roadmap for twenty-twenty eight.

00:12:45: It was running code live on the floor securely passing complex chemical and material data across the APIs of massive distinct companies like BASF, SAP & Siemens.

00:12:56: They proved secure.

00:12:57: interoperable data sharing across fierce competitors is technically solvable.

00:13:01: today

00:13:02: and pushing that interoperability down to the machine level, Tobias Lante brought attention to the beta launch of CESMI's i-IIIX.

00:13:10: This

00:13:11: is an open API specifically engineered for manufacturing interoperabilty relying on a schema first integration approach.

00:13:18: Schema First Integration?

00:13:20: Okay break that down from me because that sounds like heavy systems engineering jargon!

00:13:23: It does think of schema first integrations like adopting universal electrical outlet for your data.

00:13:29: right now factory engineers often have to hardwire every new piece.

00:13:45: He asked, Are you still connecting machines by hand?

00:13:54: Oh, that is a brutal question.

00:13:56: Because if your engineers are manually hard-coding custom data connections for every new sensor or robot... ...you will never be able to scale the AI we were just talking about.

00:14:05: You'll just be paralyzed by technical debt while your competitors use open APIs to plug and play their infrastructure.

00:14:12: But standardizing all these connections in deploying autonomous agents raises this unavoidable tension.

00:14:18: Yeah.

00:14:18: As data scales across borders and as machines begin making autonomous decisions with physical consequences, who writes the rules?

00:14:26: How does this affect global

00:14:27: competitiveness?".

00:14:29: This dominated The Halls and brings us to our final theme—policy sovereignty and

00:14:34: decarbonization.".

00:14:47: Antonio Vieira Santos and Stephen Elsham both captured a very fiery public discussion involving Siemens CEO Roland Busch, German Chancellor Merce.

00:14:57: Busch made a massive wave when he argued that it is complete nonsense to treat industrial AI like consumer AI.

00:15:03: Let's unpack the logic behind Busch' argument.

00:15:06: Why must the regulatory frameworks be separated?

00:15:09: Well,

00:15:09: because the underlying risk profiles and mechanisms of technology are entirely different.

00:15:14: If a generative AI chatbot on your phone hallucinates or recommends a movie you don't like it's minor annoyance.

00:15:20: But if an industrial AI managing high voltage power grid or robotic safety sensor hallucinate It is catastrophic failure.

00:15:28: critical infrastructure Bush's argument is that the EU AI Act is aggressively clumping these vastly different technologies together.

00:15:36: Yeah, industrial engineering Is already heavily regulated by stringent machinery and safety laws.

00:15:41: adding The broad strokes of the EU I act introduces massive regulatory drag.

00:15:45: And what does the tangible business consequence?

00:15:48: Of that regulatory friction?

00:15:49: capital flight.

00:15:51: Bush flat-out stated that because of this regulatory environment, Siemens is routing one billion euros off their AI investment away from Europe and into the United States.

00:16:00: Wow!

00:16:01: That's a profound economic indicator….

00:16:04: It highlights how digital sovereignty is no longer just political talking point for diplomats – it.

00:16:12: Europe is highly aware of this vulnerability, though.

00:16:15: Right.

00:16:15: Nicola Beer the vice president of European Investment Bank was at the fair to confirm a two point four billion euro financial package.

00:16:23: This is explicitly aimed at backing European deep tech advanced robotics and energy resilience.

00:16:29: They

00:16:30: are actively trying to finance the infrastructure that underpins strategic sovereignty in an age of intense geopolitical uncertainty.

00:16:38: And you really cannot discuss the future of industrial sovereignty without talking about energy and decarbonization.

00:16:43: To round out a global picture at Hanover, Brazil was official partner country this year.

00:16:47: Yeah they had massive footprint

00:16:49: Huge.

00:16:50: Camilla Autas posted about an incredible project discussed directly by Brazilian President Luiz Inácio Lula de Silva and Chancellor Mers.

00:16:58: They tested a European Mercedes-Benz Actros heavy duty truck, but instead of running it on standard diesel It operated on one hundred percent Bay eight Bevant biofuel which is produced in Brazil.

00:17:10: Interesting what is the mechanism there?

00:17:12: Do you have to completely reengineer their trucks powertrain to handle that?

00:17:16: That's best part.

00:17:17: no It is a drop in fuel.

00:17:19: It required only minor software parameter adaptations to the engine management system, mostly adjusting fuel injection timing.

00:17:26: yet it achieved an ninety-nine percent reduction and co two equivalent emissions.

00:17:30: Ninety nine percent that's incredible

00:17:32: right?

00:17:32: It was a massive proof of concept that sustainable industrial technology isn't some distant promise waiting for break through in twenty forty.

00:17:39: The mechanisms are mature And the solutions are ready to scale Right now.

00:17:42: yeah...it really underscores how global partnerships like bridging Brazilian biofuel resources with precision German engineering are accelerating the path to climate neutral operations.

00:17:52: It all circles back to execution reality, which has really been a heartbeat.

00:17:57: of this entire deep dive The industry is forcefully moved out of the laboratory and onto factory floor.

00:18:03: Absolutely

00:18:04: So.

00:18:04: what does it mean for you as youth listener when heading into your operation?

00:18:08: I want leave you at final thought to mull over If AI is rapidly evolving from a helpful, chatty assistant into an autonomous agent.

00:18:20: An agent that can write its own PLC code dynamically perceive it's physical environment and make split second operational decisions.

00:18:29: how will the definition of human expertise change over the next five years?

00:18:32: Oh!

00:18:32: That's great question

00:18:34: because industrial professional in future won't be managing machines.

00:18:37: They will be managing ecosystems of intelligence.

00:18:40: That is a wild paradigm shift to wrap our heads around, it really means we aren't just figuring out how to mass produce the concept car anymore.

00:18:46: We're having to entirely redefine what it means To Be The Driver.

00:18:50: Well if you enjoyed this episode.

00:18:52: new episodes drop every two weeks.

00:18:54: Also check out other editions on smart manufacturing digital construction and digital power tools.

00:18:59: Thank You for joining us On This Deep Dive.

00:19:01: Keep watching the status quo And don't forget to subscribe.

00:19:04: Catch ya next time.